Research And Application On Improved Artificial Fish Swarm Algorithm

The Artificial Fish Swarm Algorithm is a new optimization algorithm base on group intelligent that derived from fish in nature, which provide an effective tool for distributed computing for a large number of engineering problems. This algorithm has been widely used in various fields of the national economy that it does not rely on the mathematical properties of the problem,has a good robustness performance and is not sensitive to the initial values, which developed from the solution in continous optimization problems to the solution in combinatorial problems and unidimensional static state optimization problems to the solution in multi-dimensional dynamic combination optimization problems, the Artificial Fish Swarm Algorithm has become a hot research subject in optimization technical fields.The Artificial fish do a Top-down search method search mechanism. The main behavior of artificial fish's foraging behavior, cluster behavior and following behavior in imitation of fish then to achieve the global optimization. Affected by crowding factor that Artificial fish can only be hovering in the extreme neighborhood When approaches the extremum, it make artificial fish can't get the exact solution. According to the basic Particle Swarm have some capacity of tropism and rapid convergence that it can help to make up for weaknesses in the capacities of local search and low convergence speed in the later period of the optimization, so the PSO-AFSA is proposed. The algorithm is optimized on the mathematical model in artificial fish's foraging behavior, cluster behavior and following behavior. Moreover, a nonlinear dynamic inertia weight strategy and flight strategy is proposed based on the standard particle swarm algorithm. It tests the performance of algorithm through Sphere function, Ackley function, Levy function and Griewank function. Generally contrasts and researches it through many aspects such as the speed of algorithm iteration, the accuracy of algorithm convergence and the nonlinear dynamic inertia weight of algorithm. Finally, compare with traditional AFSA and PSO by some data, we proved that the PSO-AFSA has better convergence performance.TSP problem is a typical NP difficult problem. It is of great theoretic and practically significant to research on traveling salesman problems with Modern Intelligence Algorithms. TSP is short for'travelling saleman problem', which is the shortest distance by travelling saleman pass by N cities and return back iff one time. Aiming at this problem, this paper presents a cross mutation artificial fish swarm algorithm, its improved algorithm is used to solve a Traveling Salesman Problem. The crossover mutation operator is introduced based on the artificial fish swarm algorithm and definitions of distance, neighborhood and center were introduced to the solving of traveling salesman problems. Reference database of 51 city as an example, the algorithm realize the optimization of the shortest path. It shows that algorithm in solving combinatorial optimization problems, showed strong searching ability and good performance.The selection of kernel function type, kernel parameter value and error penalty parameter value directly affects the recognition effect of cancer cells recognition based on support vector machine. However, there is no scientific method to select these three factors and people select them only according to experience and repeated experiments. There exists great limitation. In this paper, we use the improved algorithm to optimize the kernel parameter and penalty factor that get a satisfactory effect in the recognition of cancer cells.