Design optimization of an artificial lateral line system incorporating flow and sensor uncertainties. Issue 2 (1st February 2017)

• Record Type: Journal Article
• Authors: Ahrari, Ali; Lei, Hong; Sharif, Montassar Aidi; Deb, Kalyanmoy; Tan, Xiaobo
• Is Part Of: Engineering optimization. Volume 49:Issue 2(2017)
• Page Start: 328
• Page End: 344
• Publication Date: 2017-02-01
• Subjects: uncertainty formulation -- dipole source localization -- bi-level optimization -- sensor placement; Engineering design -- Periodicals; Mathematical optimization -- Periodicals; 620.0042
• DOI: 10.1080/0305215X.2016.1168108
• Languages: English
• ISSNs: 0305-215X

Abstract

ABSTRACT: An artificial lateral line (ALL) system consists of a set of flow sensors around a fish-like body. An ALL system aims to identify surrounding moving objects, a common example of which is a vibrating sphere, called a dipole. Accurate identification of a vibrating dipole is a challenging task because of the presence of different types of uncertainty in measurements or in the underlying flow model. Proper selection of design parameters of the ALL system, including the shape, size, number and location of the sensors, can highly influence the identification accuracy. This study aims to find such an optimum design by developing a specialized bi-level optimization methodology. It identifies and simulates different sources of uncertainty in the problem formulation. A parametric fitness function addresses computational and practical goals and encompasses the effect of different sources of uncertainty. It can also analyse the trade-off between localization accuracy and the number of sensors. Comparison of the results for different extents of uncertainty reveals that the optimized design strongly depends on the amount of uncertainty as well as the number of sensors. Consequently, these factors must be considered in the design of an ALL system. Another highlight of the proposed bi-level optimization methodology is that it is generic and can be readily extended to solve other noisy and nested optimization problems.