Cuckoo Search-Based Least Squares Support Vector Machine Models for Optimum Tuning of Tuned Mass Dampers

Abstract

Tuned mass dampers (TMDs) have been widely used to suppress or absorb vibration. Optimum tuning of the TMD parameters using metaheuristic algorithms demands numerous numerical analyses which is a tedious and time-consuming task. Recent advances in data processing systems have attracted great attention towards the creation of intelligent systems to evolve models in engineering applications. The present paper implements the least squares support vector machine (LS-SVM) to build up models which predict the optimum TMD parameters. The performance of the proposed models is largely dependent on the quantity and the accuracy of databases used for training the models. Therefore, a wide-range numerical tuning of the TMD system, attached to a single-degree-of freedom (SDOF) main system, is done using a novel metaheuristic algorithm, called the cuckoo search (CS), to obtain the tuning frequency and damping ratio of the TMD system for a main system subjected to three types of excitations: external white-noise force, harmonic base acceleration and white-noise base acceleration. The superior performance of the LS-SVM models in prediction of optimum TMD parameters is proved in comparison to other studies in the literature. Furthermore, it is found that the optimum TMD parameters are not influenced by the predominant frequency of the filtered white-noise excitation.