Initialization Procedure (initialization + procedure)

Distribution by Scientific Domains


Selected Abstracts


Exact initialization of the recursive least-squares algorithm

INTERNATIONAL JOURNAL OF ADAPTIVE CONTROL AND SIGNAL PROCESSING, Issue 3 2002
Petre Stoica
Abstract We present an initialization procedure for the recursive least-squares (RLS) algorithm that has almost the same form as the RLS algorithm itself and which is exact in the sense that the so-initialized RLS estimate coincides with the batch LS estimate as soon as the latter exists. Copyright 2002 John Wiley & Sons, Ltd. [source]


How to optimize the TS-fuzzy knowledge base to achieve desired performances: Accuracy and robustness

OPTIMAL CONTROL APPLICATIONS AND METHODS, Issue 1 2008
A. Soukkou
Abstract Designing an effective criterion/learning to find the best rule and optimal structure is a major problem in the design process of fuzzy neural controller. In this paper, we introduce a new robust model of Takagi Sugeno fuzzy logic controller. A hybrid learning algorithm, called hybrid approach to fuzzy supervised learning (HAFSL), which combines the genetic algorithm (GA) and gradient descent technique (GD) is proposed for constructing an efficient and robust fuzzy neural network controller (FNNC). Two phases of design and learning process are presented in this work. A GA is used for finding near optimal structure/parameters of the FNNC that minimizes the number of rules (initialization procedure). The second stage of learning algorithm uses the backpropagation algorithm based on GD method to fine tune the consequent parameters of the controller. The genes of chromosome are arranged into two parts, the first part contains the control genes (the certainty factors) and the second part contains the parameters genes that representing the fuzzy knowledge base. The effectiveness of this chromosome formulation enables the fuzzy sets and rules to be optimally reduced. The performances of the HAFSL are compared to these found by the traditional PI with genetic optimization (GA-PI). Simulations demonstrate that the proposed HAFSL and GA-PI algorithms have good generalization capabilities and robustness on the water bath temperature control system. Copyright 2007 John Wiley & Sons, Ltd. [source]


Data assimilation of high-density observations.

THE QUARTERLY JOURNAL OF THE ROYAL METEOROLOGICAL SOCIETY, Issue 605 2005
I: Impact on initial conditions for the MAP/SOP IOP2b
Abstract An attempt is made to evaluate the impact of the data assimilation of high-frequency data on the initial conditions. The data assimilation of all the data available on the Mesoscale Alpine Program archive for a test case is performed using the objective analysis and the Variational Data Assimilation (Var) techniques. The objective analysis is performed using two different schemes: Cressman and multiquadric; 3D-Var is used for the variational analysis. The European Centre for Medium-Range Weather Forecasts analyses are used as first guess, and they are blended together with the observations to generate an improved set of mesoscale initial and boundary conditions for the Intensive Observing Period 2b (17,21 September 1999). A few experiments are performed using the initialization procedure of MM5, the mesoscale model from Penn State University/National Center for Atmospheric Research. The comparison between improved initial conditions and observations shows: (i) the assimilation of the surface and upper-air data has a large positive impact on the initial conditions depending on the technique used for the objective analysis; (ii) a large decrease of the error for the meridional component of the wind V at the initial time is found, if assimilation of three-hourly data is performed by objective analysis; (iii) a comparable improvement of the initial conditions with respect to the objective analysis is found if 3D-Var is used, but a large error is obtained for the V component of the wind. Copyright 2005 Royal Meteorological Society [source]


Adaptive critic design using non-linear network structures

INTERNATIONAL JOURNAL OF ADAPTIVE CONTROL AND SIGNAL PROCESSING, Issue 6 2003
Ognjen Kuljaca
Abstract A neural net (NN)/fuzzy logic (FL) adaptive critic controller is described. This structure takes advantage of the decision-making properties of a FL system to critique and tune a NN action-generating network. The stability of the proposed structure is proven. NN and fuzzy weight tuning algorithms are given that do not require complicated initialization procedures or any off-line learning phase. Tracking and bounded NN weights and control signals are guaranteed. The adaptive fuzzy critic controller given here is a model-free controller' in the sense that it works for any system in a prescribed class without the need for extensive modeling and preliminary analysis to find a regression matrix'. There is no linearity in the parameter (LIP) requirement. Copyright 2003 John Wiley & Sons, Ltd. [source]