Convergence Analysis of the MISO-OOSRIVCF Algorithm for Fractional System Identification

Continuous-time system identification with fractional model has increased in interest due to its relevance in modeling physical processes with memory and diffusion effects. This paper deals with continuous-time system identification of multiple-input single-output (MISO) fractional differentiation models. Recent studies, which have extended classic methods (e.g., OE and SRIVC) to the estimation of fractional-order models, the definition of structured-commensurability has been introduced to better cope with the estimation of fractional differentiation orders, and notably to improve the algorithm convergence. This paper investigates the convergence properties of the MISO-OOSRIVCF algorithm. After a brief method review and a SISO illustration, a detailed MISO case study is presented with convergence analysis and performance metrics. The impact of initial conditions, noise, and tuning parameters are assessed and practical strategies are then proposed for algorithm configuration for reliable system identification.