Advanced Research Of Central Force Optimization Algorithm

In reality, many problems can be abstracted into optimization problems. Consequently,people keep continuing to seek optimization algorithms. In recent years, through simulatingmechanisms and phenomena of the natural and biology, the scholars have designed manyheuristic algorithms. With the advantage of its characteristics, such as simple principles, easy toimplement, fast convergence, high accuracy, robustness and so on, people pay more and moreattention on it, which has become a very popular research direction.2007, R.A.Formatoproposed a new kind of heuristic algorithm based on gravitational particle kinematic rules—Center Force Optimization algorithm. This article mainly aims at the improvement of the centerforce optimization algorithm, the main work follows:1. Explains the principles of Central Force Optimization algorithm and its equations andprocess. Details the advantages and disadvantages of CFO algorithm.2. The Adaptive Central Force Optimization algorithm is proposed in order to avoid CFO’sunstable results. The particles fitness functions is defined, the particle movement time whichautomatic regulates with the iteration is defined as adaptative number compared with the averagefitness value, to balance the abilities of global detective and local search. The current position isupdated by the crossover operation. Using8classic benchmark functions to test, simulationresults show that, comparing with several other algorithms, ACFO has strong robustness, muchbetter than any other algorithms.3. In order to avoid obtaining local optimal solution of Central Force Optimization algorith-m, reinforcing the ability of searching, a new algorithm is proposed based on Differential Evolut-ion algorithm. According to the characteristics of Differential Evolution algorithm, differential e-volution operator mutation is introduced to mutate the component of particle and to reduce thepossibility of trapping in the local optimum and to improve the convergent speed of global searc-hing. Using6classic benchmark functions to test, simulation results show that, comparing withseveral other algorithms, the precision of the new algorithm is remarkably improved, therefore t-he effectiveness and feasibility of the algorithm is proved correctly.