A study on finite-time particle swarm optimization as a system identification method
- PDF / 692,330 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 85 Downloads / 184 Views
(0123456789().,-volV)(0123456789(). ,- volV)
TECHNICAL PAPER
A study on finite-time particle swarm optimization as a system identification method Manuel A. Ferna´ndez1 • Jen-Yuan Chang1 Received: 27 October 2020 / Accepted: 10 November 2020 Ó Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract The performance as a system identification technique of a variant of the particle swarm optimization (PSO) algorithm named finite-time particle swarm optimization (FPSO) was studied. First, this method was compared to several system identification algorithms by using data from a simulated linear system model. Special attention was given to their performance when the data from which they estimate the parameters of the system contain measurement noise. Afterwards, the effectiveness of FPSO in estimating the parameters of nonlinear systems was evaluated, using both data from simulations and data obtained from a real system with nonlinear behavior. The FPSO algorithm showed excellent performance when estimating the parameters of the simulated linear and nonlinear systems, both with noisy and noiseless data. Results from the parameter estimation of the real system showed more variation in the results of the algorithm; however, simulations using the estimated parameters were still able to closely emulate the real system.
1 Introduction Inspiration for this study came when our research had the necessity to know the parameters of a certain DC motor system which contained a dry friction nonlinearity. Parameters from such a system can be estimated using a system identification method. This type of estimation method is used to estimate the parameters of a dynamical system whose structure is known, but whose parameters might not be easily measured. They make use of knowledge of the structure of a dynamical system, as well as of measurements of the input and output quantities of the system, to estimate the parameters corresponding to this system. Among several system identification methods, Finite-time Particle Swarm Optimization (FPSO) seemed to be the most appropriate for the DC motor parameter estimation task. FPSO, first presented by Lu and Han (2012), is a variant of the Particle Swarm Optimization algorithm (PSO) (Kennedy and Eberhart 1995), a type of optimization algorithm which can be used as a system & Jen-Yuan Chang [email protected] Manuel A. Ferna´ndez [email protected] 1
Department of Power Mechanical Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan
identification technique. This type of algorithm uses a swarm of particles to find the minimum of a cost function. It was then desired to have a deeper understanding of how such an estimation method compares to other system identification methods, as well as to know how well it performs at estimating parameters for linear and nonlinear systems while using noisy data. Based on this desire of a deeper understanding of FPSO, this study was carried out with the objectives of discovering how suitable is the FPSO algorithm to be used
Data Loading...
