Continuous-Time Model Identification Using Reinitialized Partial Moments - Application to Power Amplifier Modeling

Djamai, Mourad; Tohme, Elie; Ouvrard, Régis; Bachir, Smaïl

doi:10.3182/20060329-3-au-2901.00205

Cited by 13 publications

(13 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…They are obtained for an input signal with low input level and large bandwidth. Parameter estimation is performed by iterative Instrumental Variable based on Reinitilized Partial Moments RPM 1 method (see also [10]). …”

Section: B Initialization Problemsmentioning

confidence: 99%

“…Moreover, OE methods can be used to identify nonlinear systems. For these advantages, the OE methods are more appropriate in microwave systems characterization [10]. For PA identification, the parameters initialization and input excitation are very important to ensure global convergence.…”

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…On the opposite, continuous model is a system where the mathematical representation, under some assumptions, is perfectly known. The drawback of this model is the complexity of the electrical modelling but the main advantage is that the resulting parameters have physical significance like gain conversion, damping coefficient and cut-off frequency [10].The model considered in this paper is a Multi-Input/MultiOutput (MIMO) system described in continuous-time domain. This structure is similar to Hammerstein discrete-time model including nonlinear transfer functions and multivariable continuous filter.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Modeling and identification of continuous-time system for RF amplfiers

Djamai

Bachir

Duvanaud

2008

2008 38th European Microwave Conference

Self Cite

View full text Add to dashboard Cite

In this paper, we present a new identification procedure for radio frequency Power Amplifier (PA) in the presence of nonlinear distortion and memory effects. The proposed procedure uses a continuous-time model where PA dynamics are modeled with a multivariable filter and a general polynomial function. Using the baseband input and output data, model parameters are obtained by an iterative identification algorithm. Finally, the proposed estimation method is tested and validated on experimental data by comparison of the quadrature IQ signals in time domain. I. IntroductionNumerous approaches in Power Amplifier identification area have been developed to characterize the input to output complex envelope relationship [1][2] [3]. The model forms used in identification are generally classified into two methods depending on the physical knowledge of the system: discrete and continuous model. A discrete model is a system where no physical insight and prior information available. This approach have been widely used in many research studies to predict and linearize the output of the Nonlinear PA such as neural networks and Volterra series [4][5][6][7]. However, this method suffer from the high number of parameters and the time consuming in computation. On the opposite, continuous model is a system where the mathematical representation, under some assumptions, is perfectly known. The drawback of this model is the complexity of the electrical modelling but the main advantage is that the resulting parameters have physical significance like gain conversion, damping coefficient and cut-off frequency [10].The model considered in this paper is a Multi-Input/MultiOutput (MIMO) system described in continuous-time domain. This structure is similar to Hammerstein discrete-time model including nonlinear transfer functions and multivariable continuous filter.Model parameters are achieved using an iterative identification algorithm based on Output Error (OE) method. This technique is based on minimization of a quadratic criterion by a Non Linear Programming (NLP) algorithm. This technique requires much more computation and do not converge to unique optimum. But, OE methods present very attractive features, because the simulation of the output model is based only

show abstract

Section: B Initialization Problemsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Modeling and identification of continuous-time system for RF amplfiers

Djamai

Bachir

Duvanaud

2008

2008 38th European Microwave Conference

Self Cite

View full text Add to dashboard Cite

show abstract

“…But the two methods have a drawback. The IDIM method needs a good tuning of filters [3] [4] [5] while the OE method needs good initial conditions [9] and a powerful algorithm to solve non linear least squares [11]. By mixing these two approaches [12], our method overcomes these drawbacks.…”

Section: Introductionmentioning

confidence: 99%

Global identification of mechanical and electrical parameters of asynchronous motor driven joint with a fast CLOE method

Robet

Gautier

2013

2013 15th European Conference on Power Electronics and Applications (EPE)

View full text Add to dashboard Cite

In many cases of new actuation of compliant controlled or bio-inspired joint driven robot, a global identification of electrical and mechanical coupled dynamics is required. This paper proposes a technique which mixes a closed loop output error method with the inverse dynamic identification model method which allows using linear least-squares technique to estimate the parameters. A first approach which has been validated on a DC motor allows a decoupled identification of the electrical and mechanical dynamics but fails to make a simultaneous identification. A major improvement of that method is proposed to carry out the coupled identification of both mechanical and electrical parameters. A validation on an asynchronous motor driven joint shows the effectiveness of the new procedure.

show abstract