Model Order Reduction with True Dominant Poles Preservation via Particles Swarm Optimization
- PDF / 674,848 Bytes
- 13 Pages / 439.37 x 666.142 pts Page_size
- 21 Downloads / 126 Views
Model Order Reduction with True Dominant Poles Preservation via Particles Swarm Optimization Othman Alsmadi1 · Adnan Al-Smadi2 · Mohammed Ma’aitah3 Received: 24 July 2019 / Revised: 28 April 2020 / Accepted: 30 April 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract A new computational technique for the reduction of multi-time scale systems is proposed in this paper. The reduction process is performed based on the dominant poles preservation in the reduced-order model. The true dominant poles are selected based on the highest contribution in redefined time moments and lowest contribution in redefined Markov parameters. Motivated by the singular perturbation approximation, obtaining the reduced-order model will be achieved by using the artificial intelligent method named particles swarm optimization. The potential of the proposed technique is observed when comparing its results with other recently published methods. Keywords Particles swarm optimization · Model order reduction · True dominant poles · Singular perturbation approximation
1 Introduction In recent years, making decisions in many different types of systems have assumed an increasingly important role in the development and advancement of modern civilization and technology [11]. This fact has attracted many researchers to focus on different optimization problems [15, 23, 24]. For analysis and design, many physical systems are represented by mathematical formulation, however with higher-order differential equations [1]. To simplify such high-order systems, the technique of model order reduction (MOR) becomes very important [3]. Working with reduced-order models
B
Othman Alsmadi [email protected] Adnan Al-Smadi [email protected]
1
Electrical Engineering Department, The University of Jordan, Amman, Jordan
2
Electronics Engineering Department, Yarmouk University, Irbid, Jordan
3
Electrical Power Engineering Department, Yarmouk University, Irbid, Jordan
Circuits, Systems, and Signal Processing
makes it easier to implement analysis, simulations, and control system designs [22]. Although a significant number of methods for MOR do exist and all aim for accurate reduced models for low computational costs, no approach always gives the best results for all systems. Each method has its advantages and disadvantages [6]. For MOR, a system may be reduced using different methods [1, 8, 12, 19]. One method obtains reduced-order models that are completely new and not related to the original models, but provides performance very similar to the original models [1]. Another method obtains reduced-order models but preserves the original system important properties, such as dominant poles which provide the system critical dynamics. It has been shown that the later approach is more preferable due to its physical consideration [1]. Numerous methods of MOR have been proposed and are available in the literature [2, 4–6, 10]. The authors in [6] proposed an MOR technique with the advantage of critical frequency preservation capability us
Data Loading...