9 skills found
baggepinnen / ControlSystemIdentification.jlSystem Identification toolbox, compatible with ControlSystems.jl
asymppdc / AsympPDCThe asympPDC Package is a MATLAB and Octave package for Partial Directed Coherence (PDC) and Directed Transfer Function (DTF) estimation with asymptotic statistics, allied functions and routines for Granger Causality Test and results pretty plotting.
chrystalchern / Mdofdynamic analysis of structural vibrations.
RuttenStijn / ThesisCode and extra figures as part of the thesis about Relative transfer function estimation for multi-microphone speech enhancement based on the Kalman filter
Screeen / SVD DirectOfficial implementation of RTF estimation for acoustic beamforming from "Wideband Relative Transfer Function (RTF) Estimation Exploiting Frequency Correlations", published in TALSP 2025
levidaniel96 / PeerRTFRobust Relative Transfer Function (RTF) Estimation using Graph Neural Networks
omriMatania / One Fault Shot Learning For Gear Severity EstimationCodes of "One-fault-shot learning for fault severity estimation of gears that addresses differences between simulation and experimental signals and transfer function effect"
Ahmed-ElTahan / Stochastic Recursive Modified Extended Least Squreas With Exponential Forgetting Factor RMELSWEF% This function is made by Ahmed ElTahan %{ This function is intended to estimate the parameters of a dynamic system of unknown parameters using the Recursive Modified Extended Least Squares With Exponential Forgetting Factor Method (RMELSWEF) for time varying parameter system which has an noise addition. After an experiment, we get the inputs, the outputs of the system. The experiment is operated with sample time Ts seconds. The model is given by A(z) y(t) = B(z)sys u(t) + C(z) eps(t) which can be written in z^(-d) B(z) C(z) y(t) = ------------------- u + ------------ e = L*u + M*e A(z) A(z) where: -- y : output of the system. -- u : control action (input to the system). -- e : color guassian noise (noise with non zero mean). -- Asys = 1 + a_1 z^-1 + a_2 z^-2 + ... + a_na z^(-na). [denominator polynomail] -- Bsys = b_0 + b_1 z^-1 + b_2 z^-2 + ... + b_nb z^(-nb). [numerator polynomail] -- C = 1 + c_1 z^-1 + c_2 z^-2 + ... + c_nc z^(-nc). [noise characteristics] -- d : delay in the system. A and C are monic polynomials. (in output estimation of the stochastic system as C is monic, we add e(t) to the estimation i.e. not starting from c1*e(t-1)) Function inputs u : input to the system in column vector form y : input of the system in column vector form na : order of the denominator polynomail nb : order of the numerator polynomail nc : order of the characteristics of the noise (usually <=2 for max) d : number represents the delay between the input and the output lambda : forgetting factor -->>> 1>lambda>0 Function Output Theta_final : final estimated parameters. Gz_estm : pulse (discrete) transfer function of the estimated parameters 1 figure for the history of the parameters that are being estimated 2 figure to validate the estimated parameters on the given output using the instantaneous estimated parameters. 3 figure to plot the input versus time. Note: the noise added shall not to be with a magnitude close to the system output, it should be smaller, this is in simulation such as here or the algorithm will go crazy that can't distinguish between the main and the noisy signal (This can be measured in practical case finding noise to signal ratio). An example is added to illustrate how to use the funcrtion %}
Ahmed-ElTahan / Stochastic Recursive Extended Least Squreas With Exponential Forgetting Factor RELSWEF% This function is made by Ahmed ElTahan %{ This function is intended to estimate the parameters of a dynamic system of unknown parameters using the Recursive Extended Least Squares With Exponential Forgetting Factor Method (RELSWEF) for time varying parameter system. After an experiment, we get the inputs, the outputs of the system. The experiment is operated with sample time Ts seconds. The model is given by A(z) y(t) = B(z)sys u(t) + C(z) eps(t) which can be written in z^(-d) B(z) C(z) y(t) = ------------------- u + ------------ e = L*u + M*e A(z) A(z) where: -- y : output of the system. -- u : control action (input to the system). -- e : color guassian noise (noise with non zero mean). -- Asys = 1 + a_1 z^-1 + a_2 z^-2 + ... + a_na z^(-na). [denominator polynomail] -- Bsys = b_0 + b_1 z^-1 + b_2 z^-2 + ... + b_nb z^(-nb). [numerator polynomail] -- C = 1 + c_1 z^-1 + c_2 z^-2 + ... + c_nc z^(-nc). [noise characteristics] -- d : delay in the system. A and C are monic polynomials. (in output estimation of the stochastic system as C is monic, we add e(t) to the estimation i.e. not starting from c1*e(t-1)) Function inputs u : input to the system in column vector form y : input of the system in column vector form na : order of the denominator polynomail nb : order of the numerator polynomail nc : order of the characteristics of the noise (usually <=2 for max) d : number represents the delay between the input and the output lambda : forgetting factor -->>> 1>lambda>0 Function Output Theta_final : final estimated parameters. Gz_estm : pulse (discrete) transfer function of the estimated parameters 1 figure for the history of the parameters that are being estimated 2 figure to validate the estimated parameters on the given output using the instantaneous estimated parameters. 3 figure to plot the input versus time. Note: the noise added shall not to be with a magnitude close to the system output, it should be smaller, this is in simulation such as here or the algorithm will go crazy that can't distinguish between the main and the noisy signal (This can be measured in practical case finding noise to signal ratio). An example is added to illustrate how to use the funcrtion
