Freestyle Rap Harmony Search (FRHS) for Engineering Problem Optimization

摘要

A maximum benefit with minimum cost has been pursued in the optimal design of water distribution systems (WDSs). For this problem, trial-and-error methods, mathematical approaches, dynamic programming, and meta-heuristic techniques have been proposed while satisfying all the hydraulic constraints such as minimum pressure and demand. Recently, various variants of HSA mostly with multiple harmony memories and hierarchical memories have drawn a lot of attention because of their promising search. In this study, we propose a new version of HSA, freestyle rap harmony search (FRHS), with multiple memories among which different types of solutions are shuffled. The core properties of Freestyle rap are Rhyme, Flow, and Punch line. In FRHS algorithm, multiple Harmony Memories (HMs) are created as flow property in freestyle rap. When the value of the number of solution vector generations (NSVG) is above a certain value as a parameter, the bad solution of each HM is replaced by the good solution of the neighboring HM. The movement of the harmony memory vector is like a rhyme property with similar pronunciation repeated. When NSVG is above a certain value as a parameter, this condition is the same as the property of the punch line, the most shocking part of the freestyle rap. First, to check the performance of the algorithm, FRHS is applied to the least-cost design of WDSs. The performance of FRHS and HS is evaluated by how fast to find nearly global optimal solution. FRHS converges to the nearly global optimal solution faster than HS.