Forgetting Factor Approach Research Articles

Sliding Mode Observers for Plasma Signal Identification in Remote Laser Welding

In this paper, a model for the emission of a plasma signal during the zinc coated steel metal fusion phenomenon is proposed. Remote laser welding process and system identification of plasma signal based on the sliding mode observer are presented. The dynamic behaviour of the keyhole is captured using experimental data from the photodiode sensor. A multi-input single-output (MISO) model is designed, the equivalent injection signal using a unit vector sliding mode observer is applied to update the dynamic matrix coefficients. The variation in the model parameters due to process disturbances in the laser fusion phenomenon is captured by recursively updating the matrix parameters, using matrix forgetting factor approach. The simulation result shows an accurate estimation of plasma signal emitted during the laser welding process in the presence of process disturbances.

Simulation of Frequency Spectrum of Electric Power Signal

In this paper, recursive least squares method (RLSM) which is one of the adaptable classical methods is used. Forgetting factor approach is adapted to RLSM to obtain phase information of voltage signal belonging to an electric power network at first. Then responses of the algorithm are investigated for the signal which has zero magnitude for a specific time interval. And the examination of the frequency spectrum with RLSM method algorithm is done. Simulations are carried out by developing MATLABTM codes. Simulation results of method are presented. Ill. 4, bibl. 9 (in English; abstracts in English, Russian and Lithuanian).

Reservoir Inflow Modeling Using Temporal Neural Networks with Forgetting Factor Approach

In this paper, a recursive training procedure with forgetting factor is proposed for on-line calibration of temporal neural networks. The forgetting factor discounts old measurements through an on-line model calibration. The forgetting factor approach enables the recursive algorithm to reduce the effect of the older error data by multiplying the error data by a discounting factor. The proposed procedure is used to calibrate a temporal neural network for reservoir inflow modeling. The mean monthly inflow of the Karoon-III reservoir dam in the south-western part of Iran is used to test the performance of the proposed approach. An autoregressive moving average (ARMA) model is also applied to the same data. The temporal neural network, which is trained with the proposed approach, has shown a significant improvement in the forecast accuracy in comparison with the network trained by the conventional method. It is also demonstrated that the neural network trained with forgetting factor results in better forecasts compared to the statistical ARMA model, which has been calibrated through this approach.

A New Approach of Kernel Bandwidth Applications for Time-series Using the Example of the Prediction of the Euro's Exchange Rate

In recent years the application of kernel smoothing methods in nonparametric regression framework to financial time-series analysis has become widespread. Kernel smoothing methods have not been applied, however, to a wide range of problems arising in time-series simulations and forecasting. Forgetting factors can be both fixed and variable. Gijbels et al (1999) propose an understanding of fixed forgetting factors via kernel smoothing. However the variable forgetting approach is not mentioned. This paper describes the variable forgetting factor and the fixed forgetting factor, and establishes the linkage for the first time between the variable forgetting factor approach and kernel smoothing. The forgetting factor method uses a sample of data and estimates the value of the forgetting factor from the sample. This method fits better than a parametric approach, which uses some assumed parameters. Since the forgetting factor method is equivalent to a kernel estimation - which is a non-parametric method - it is likely to give more accurate estimates and better forecasting performance in financial time-series than a parametric one. The major area of interest in this application is whether kernel estimation, using Cho's approach [see Cho et al (1991) and Brailsford et al (2002)] for kernel bandwidth selection, can improve the Euro's forecasting performance within the framework of subset AR modelling. The forecasting performance is compared with the performance of AR modelling without the use of the forgetting factor. If improved forecasting performance is achieved, this can increase the potential use of kernel smoothing methods in time-series forecasting. The findings show that the kernel bandwidth so determined can improve the forecasting performance.

On‐line identification and damage detection in non‐linear structural systems using a variable forgetting factor approach

AbstractThis paper presents an adaptive on‐line identification algorithm based on a newly defined variable forgetting factor approach. At each time step, the proposed methodology, a recursive least‐square (Kalman filter) based algorithm, upgrades the adaptation gain matrix using an adaptive forgetting factor that is expressed as the ratio between the minimum value of the diagonal elements of the adaptation gain matrix and a set of pre‐defined threshold values. This approach requires only acceleration measurements and is particularly robust to the integration errors introduced in determining velocities and displacements. Such an algorithm has been implemented for both parametric and non‐parametric identification studies. For the case of no a priori information on the type of the structural model, an analytical method based on a power series of multivariable polynomial expansions has been introduced. The effectiveness and robustness of the proposed approach has been shown in various numerical simulations, considering the effects of excitation amplitude, of measurement noise, of a priori mass estimation error, and of exact‐, under‐, and over‐parameterization of the structural model. General polynomial‐type non‐linear models such as Duffing and Van der Pole oscillators and Bouc–Wen hysteretic model have been analyzed using the proposed modeling process. The methodology presented in this paper leads to a more accurate and controllable on‐line identification either for the estimation of time‐invariant or time‐varying structural parameters (for damage detection purposes) or for the prediction of the future structural response to dynamic loading. Copyright © 2003 John Wiley & Sons, Ltd.

On-line state prediction of engines based on flat neural network

A flat neural network is designed for the on-line state prediction of engine. To reduce the computational cost of weight matrix, a fast recursive algorithm is derived according to the pseudoinverse formula of a partition matrix. Furthermore, the forgetting factor approach is introduced to improve predictive accuracy and robustness of the model. The experiment results indicate that the improved neural network is of good accuracy and strong robustness in prediction, and can apply for the on-line prediction of nonlinear multi input multi output systems like vehicle engines.

A STABLE ADAPTIVE ALGORITHM FOR LINEAR SYSTEMS PARAMETER IDENTIFICATION

This paper develops an updating algorithm for identifying the parameters of a known structure linear system. As it is known that recursive least squares algorithms are the best for identifying the parameters of time invariant systems. However, in case of having time varying system, the above algorithm can not track the variation in parameters in a good shape. Therefore, two approaches have been developed in the literature to resolve this problem i.e., the forgetting factor approach and the finite window one. This paper unifies the above two approaches in a stable fashion by having an algorithm that contains a variable forgetting factor and a constant window. The effectiveness of the proposed algorithm in tracking the system variable parameters is demonstrated by an example.