For some classes of nonlinear systems or time series, an operating point dependent ARMA model can be used to represent the system. In this paper we use the neural networks to identify such a model which can then be converted to its equivalent statespace re
Modelling and Adaptive Filtering of Nonlinear Systems Using Neural Network Zhi Qiao Wu and Chris J. Harris
Image, Speech and Intelligent System Research Group Department of Electronics and Computer Science University of Southampton, Southampton SO17 1BJ, UK Email: zqw@ecs.soton.ac.uk http://www-isis.ecs.soton.ac.uk Technical Report: c University of Southampton
For some classes of nonlinear systems or time series, an operating point dependent ARMA model can be used to represent the system. In this paper we use the neural networks to identify such a model which can then be converted to its equivalent statespace representation. Using this state-space form, a Kalman lter can be applied to estimate the state, and a simulated example is given. Keywords: Nonlinear System, System Identi cation, State Estimation, Kalman Filter.
Abstract
1 Introduction The general form of a discrete-time, single-input, single-output (SISO) stochastic nonlinear system can be given by:
y(t)= f (y(t? 1);:::; y(t? n); u(t? d? 1);:::; u(t? d? m))+ !(t);
(1)
where f (:) is an unknown nonlinear function, u(t) and y(t) are the measured input and output of the system respectively, and m; n and d are known a priori and represent the orders and time-delays of the model, f!(t)g denotes a white noise sequence. For some classes of nonlinear systems, if a linear model at a xed operating point is used for a local system representation, a global system representation can be realised via a composite set of linear models which correspond to the di erent operating points (the total number of linear models may be nite or in nite). It is therefore possible to express certain global nonlinear systems as a set of linear systems whose parameters are unknown nonlinear functions of its measurable operating points. In this case, the non-linear system in a general model (1) can then be expressed as:
y(t)= a (Ot )y(t? 1)++ an(Ot)y(t? n)+b (Ot)u(t? d? 1)++ bm (Ot)u(t? d? m)+ !(t); 1 1
(2)
Adaptive non-linear compensation control based on neural networks for non-...requires the accurate system models and is sensitive to the modelling errors...
Modelling and Adaptive F... 暂无评价 15页 ...Nonlinear control of a heating, ventilating,and ...of HVAC systems using a dynamic neural network. ...
filters represent a general nonlinear filtering ...Volterra filters, like neural network based filters...Modelling of nonlinear systems. Noise and echo ...
Nonlinear Dynamic Modelling Using Fully Tuned RBF ...Neural Networks Direct Adaptive Control for a ...filtering for linear systems with Markovian jump ...
The use of an axisymmetric quasi-steady flow ...Secondly, due to high nonlinearities, the ...modelling The present paper is devoted to ...
Modelling and Adaptive F... 暂无评价 15页 ... algorithm for nonlinear adaptive filtering using...The system output dn was corrupted by an ...
A Convolutional Neural Network for Modelling ... a bias b ∈ Rd and a nonlinear function g ...2011. Adaptive subgradient methods for online ...
Kalman Filter - filtering, smoothing and parameter... of control systems based on neural networks ? ...RRMT - rainfall-runoff modelling ? SAM - ...
filtering, smoothing and parameter estimation (using...systems based on neural networks NNSYSID - neural...modelling SAM - structure and motion Schwarz-...
filtering, smoothing and parameter estimation (using...systems based on neural networks NNSYSID - neural...modelling SAM - structure and motion Schwarz-...
Different Approaches of Artificial Neural Networks ...system: Function Fitting (FitNet), Nonlinear ...to the adaptive filtering problem based on LMSE....
(17) Soft Sensing Modelling Based on Optimal ...(19)Kalman Filtering with Partial Markovian Packet...systems; Neural networks; Nonlinear systems; ...
我要评论