The Research Of Cat Swarm Optimization Algorithm And Its Improved Algorithms

A new swarm intelligence algorithm called cat swarm optimization (CSO) which imitates the natural behavior of cats has proposed. Its simple concept, high speed of convergence and strong stability make CSO draw more and more attention. CSO has been successfully applied to functions optimization, image processing and networks training, etc. However, both its theory and application are still far from mature and are waiting for further study.At first, the research background and the fundamental recursive equation and procedure of CSO are reviewed. The connections and differences between CSO and PSO are analyzed. We present an analysis of the convergence behavior and swarm diversity of CSO. How to improve the performance of CSO are researched. The main contributions and innovations of this dissertation can be summarized as follows:1. A fractional-order cat swarm optimization optimization (FCSO) is proposed. Considering the deficiencies of inertia weight of CSO, A new form of the higer fractional-order inertia weight in the tracing mode of algorithm is used, and the principles of setting parameters are given through the experiments. FCSO is tested for several well-known functions and the comparison indicates that FCSO improves the performance of its final solution, its convergence speed and lower the probability that the population trapped in a local optimum.2. A vibrational mutation cat swarm optimization (VMCSO) is proposed.After analyze the velocity update equation and position update equation in an analytical way, we can found that the path of cat can be considered as an oscillation phenomenon around the point of the best position. They surf on an underlying foundation of sine waves, so the drawback of the tracing mode is due to the lack of diversity in the global searching. We designed a new mutation operator called a global vibrational mutation strategy for the positions updating equation in the tracing mode. It efficiently increased diversity of the swarm in the global searches, and improved the performance of its final solution and its convergence speed.