This article focuses on developing a new approach for diagnosing of nonlinear systems. This approach consists in estimating the input fault by using the proposed ARX-Laguerre multimodel (Adaily, 2018) in its decoupled structure. This method is processed in two consecutive steps. The first consists in the offline identification of parameters of the ARX-Laguerre multimodel (weighting functions, Laguerre poles and -Fourier coefficients). The second step proposes an input fault estimation algorithm based on the online update of the parameters using the sliding window principle. This proposed diagnosis approach is implemented on a Continuously Stirred Reactor (CSTR) Benchmark to estimate the input fault. A comparative study regarding the results provided by a Proportional Integral (PI) observer based on the decoupled ARX-Laguerre multimodel is performed to attest this new approach.
Multimodel, ARX Laguerre, Genetic Algorithm, Parameter optimisation, Sliding window, Input fault estimation, PI observer.
Hajer BENAMOR, Marwa YOUSFI, Larbi CHRIFI ALAOUI, Hassani MESSAOUD, "New Approach for Estimating the Input Fault Based on the Sliding Window Identification Technique of the SISO ARX-Laguerre Multimodel", Studies in Informatics and Control, ISSN 1220-1766, vol. 32(4), pp. 115-127, 2023. https://doi.org/10.24846/v32i4y202311
All articles published by Studies in Informatics and Control (SIC) are published in accordance with the terms and conditions of the Creative Commons Attribution License (CC BY-NC-SA 4.0).