Big Bang Search (bbs) Intelligence Optimization And Its Improvement

Optimization is an application technology which based on mathematics for searching optimal or suboptimal solutions of engineering problems. Due to the character of large-scale, strong constraint, nonlinear, multi-extreme, multi-targets and hard to model and so on, many actual engineering problems are NP-hard problems. So seeking an intelligence parallel algorithm fitting for those problems is a main research target for related disciplines and is a remarkable research area. Since 1980s, a series of modern optimization algorithms had been created and successfully applied to solve some complicated problems which made them get more and more attention of scientists and widespread application. Nearly each algorithm is inspired by some natural phenomena, abstract it using mathematics and create algorithm model, and then solving it by computer, so they are called natural heuristic or intelligence algorithms. Algorithms with the character of intelligence are widely applied in various fields, such as power system, chemical system, mechanical design, communications, economic field and so on, even they are involved in humanities and social science field.Inspired by Big Bang Theory, a new diffuse-type search method with double-parallel search, Big-Bang Search (BBS), is proposed in this thesis. BBS combined the uniformity and randomicity well of candidates solutions distributed in the search space, that reinforced its search ability. Furthermore, in order to fully utilize the information of explosive pieces of basic BBS, inspired by classical optimization theory, this thesis proposed an improved method: BBS based on approximate gradient (AGBBS). In a certain extent, AGBBS overcomes the deficiency of the BBS and improved the overall efficiency.This thesis consists of five parts. The first is an introduction about research significance and progress. The second elaborates the basic BBS algorithm, including source of inspiration, search mechanism, key processes and trait of the method. In addition, some heuristic operators are also discussed and the steps of the BBS are given. The third part elaborates the improved method, BBS based on approximate gradient (AGBBS). The fourth part is test for instance, including experimental results and its related analysis. The fifth part is conclusion including some thoughts achieved in research and the future work. The end is acknowledgement and references. Pseudocode of BBS and flow chart of AGBBS are given in appendix.