Identification of continuous-time model of hammerstein system using modified multi-verse optimizer

• Main Author: Most. Julakha, Jahan Jui
• Format: Thesis
• Language: English
• Published: 2021
• Subjects: TA Engineering (General). Civil engineering (General); TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://umpir.ump.edu.my/id/eprint/34941/1/Identification%20of%20continuous-time%20model%20of%20hammerstein%20system%20using%20modified%20multi-verse%20optimizer.ir.pdf; http://umpir.ump.edu.my/id/eprint/34941/

his thesis implements a novel nature-inspired metaheuristic optimization algorithm, namely the modified Multi-Verse Optimizer (mMVO) algorithm, to identify the continuous-time model of Hammerstein system. Multi-Verse Optimizer (MVO) is one of the most recent robust nature-inspired metaheuristic algorithm. It has been successfully implemented and used in various areas such as machine learning applications, engineering applications, network applications, parameter control, and other similar applications to solve optimization problems. However, such metaheuristics had some limitations, such as local optima problem, low searching capability and imbalance between exploration and exploitation. By considering these limitations, two modifications were made upon the conventional MVO in our proposed mMVO algorithm. Our first modification was an average design parameter updating mechanism to solve the local optima issue of the traditional MVO. The essential feature of the average design parameter updating mechanism is that it helps any trapped design parameter jump out from the local optima region and continue a new search track. The second modification is the hybridization of MVO with the Sine Cosine Algorithm (SCA) to improve the low searching capability of the conventional MVO. Hybridization aims to combine MVO and SCA algorithms advantages and minimize the disadvantages, such as low searching capability and imbalance between exploration and exploitation. In particular, the search capacity of the MVO algorithm has been improved using the sine and cosine functions of the Sine Cosine Algorithm (SCA) that will be able to balance the processes of exploration and exploitation. The mMVO based method is then used for identifying the parameters of linear and nonlinear subsystems in the Hammerstein model using the given input and output data. Note that the structure of the linear and nonlinear subsystems is assumed to be known. Moreover, a continuous-time linear subsystem is considered in this study, while there are a few methods that utilize such models. Two numerical examples and one real-world application, such as the Twin Rotor System (TRS) are used to illustrate the efficiency of the mMVO-based method. Various nonlinear subsystems such as quadratic and hyperbolic functions (sine and tangent) are used in those experiments. Numerical and experimental results are analyzed to focus on the convergence curve of the fitness function, the parameter variation index, frequency and time domain response and the Wilcoxon rank test. For the numerical identifications, three different levels of white noise variances were taken. The statistical analysis value (mean) was taken from the parameter deviation index to see how much our proposed algorithm has improved. For Example 1, the improvements are 29%, 33.15% and 36.68%, and for the noise variances, 0.01, 0.25, and 1.0 improvements can be found. For Example 2, the improvements are 39.36%, 39.61% and 66.18%, and for noise variances, the improvements are by 0.01, 0.25 and 1.0, respectively. Finally, for the real TRS application, the improvement is 7%. The numerical and experimental results also showed that both Hammerstein model subsystems are defined effectively using the mMVO-based method, particularly in quadratic output estimation error and a differentiation parameter index. The results further confirmed that the proposed mMVObased method provided better solutions than other optimization techniques, such as PSO, GWO, ALO, MVO and SCA.