Optimization;Statistics;Sociology;Standards;Convergence;Search problems;Linear programming;seed algorithm;random selection;archive;eigenvector crossover;experimental study;real-world problems

In this paper, an enhanced variant of the efficient nature-inspired Tree-seed optimisation algorithm (TSA) is designed. The original TSA lacks memory of old good solutions and mechanism for rotated objective functions. Two new mechanisms are used to increase the efficiency of the original TS algorithm. At first, archive A for successful seeds is used to help unsuccessful trees (TSrA). Secondly, rotationally invariant Eigenvector transformation of seeds is employed to cope with rotated problems (TSrAeig). Newly proposed TS variants are applied on a set of 22 real-world optimisation problems from CEC 2011. These problems enable to evaluate methods properly for future using in real applications. Results achieved by TSrA and TSrAeig are compared with results of the original TSA algorithm. Newly proposed mechanisms increase the performance of the original TSA algorithm significantly.