شناسايي و حذف اختلال هاي سينوسي ميرا با استفاده از فيلتر كالمن توسعه يافته

چكيده انگليسي :

77 Identification and Rejection of Exponentially Damped Sinusoidal Signal Using Extended Kalman Filter Mehrdad Yazdanian mehr yazdanian@gmail com March 04 2012 Department of Electrical and Computer Engineering Isfahan University of Technology 84156 83111 IranDegree M Sc Language FarsiSupervisors Dr Mohsen Mojiri mohsen mojiri@cc iut ac ir Dr Farid Sheikholeslam sheikh@cc iut ac irAbstract In this thesis parameter estimation of a signal consisting of a number of exponentially damped sinusoidalsignal EDS corrupted with noise is considered It is clear that this signal with zero damping factor is a puresinusoidal signal Frequency estimation of a pure sinusoidal signal is frequently encountered in signalprocessing and control applications for instance in active noise control power systems computer hard diskdrive harmonic control rotating mechanical systems etc In order to estimate the frequency of a puresinusoidal signal many techniques have been developed such as adaptive notch filter discrete Fouriertransform DFT and its modifications phase locked loop and Kalman filtering In the case of non zero damping factor both damping factor and the frequency of the EDS signal shouldbe estimated simultaneously This kind of problem arises in several practical fields such as speech audioanalysis linear system identification and transient analysis In these applications the observed signals can bemodelled as noisy EDS signals On the other hand EDS signals are frequently encountered in the response oflinear and time invariant systems Therefore the problem of estimating the parameters of an EDS signal innoise has been an attractive one although its inherent nonlinearity has made it a difficult problem Various methods have been proposed to estimate the parameters of an EDS signal Some of well knownapproaches include matrix pencil method estimation of signal parameters via rotational invariance technique ESPRIT Kung s algorithm maximum likelihood method linear prediction methods such as Prony smethod Kumaresan Tufts method These methods are not suitable for tracking time varying parameters Therefore researches have been focused on applying on line estimators such as Kalman filter to track timevarying frequency and damping factor Kalman filter is considered as a recursive algorithm for optimal estimation of linear stochastic processes This algorithm is suitable for applying in digital systems and is a reasonable choice for on line applicationsbecause of its acceptable computational load Extended Kalman filter represents the conversion of theKalman filter to nonlinear processes such as frequency tracking problem There is a long history of applyingextended Kalman filter EKF for estimation and tracking the frequency of a pure sinusoidal signal The frequency estimation based on EKF is particularly pertinent in view of our interest in developing anEDS signal parameters estimator In this thesis is we drive an EKF for estimating the parameters of an EDSsignal First a new state space model is developed for an EDS signal Then an EKF is designed using thelinearization of the proposed state space model that provides a direct estimation of frequency and dampingfactor and stability analysis of the proposed algorithm is carried out Adaptive nature of the proposed EKFenables it to track time variations of EDS signal parameters Various simulations show the desirableperformance of the proposed EKF Furthermore the proposed method is extended for a signal consisting of anumber of exponentially damped sinusoidal signals and finally the estimated parameters are employed in anadaptive feedforward cancellation scheme to cancel the EDS type disturbances affecting a linear system Keywords parameter estimation exponentially damped sinusoidal signal EDS extendedKalman filter EKF disturbance rejection

يزدانيان، مهرداد

شناسايي و حذف اختلال هاي سينوسي ميرا با استفاده از فيلتر كالمن توسعه يافته

تخمين فركانس , سيگنال سينوسي ميرا , حذف اغتشاش