Complex System Evaluation Based On Swarm Optimization Algorithm

System evaluation lies in the position of system engineering theory and methodology, which is the hot spot and difficulty of research on system engineering theory and practice. There are many problems in application such as high-dimensional nonlinear optimization problem because of its complexity, so we need introduce new intelligent method in system evaluation. Currently particle optimization algorithm has gained wide application in the function optimization, neural network training, combinatorial optimization, etc. This article attempts to applied particle optimization algorithm to complex system evaluation.On the basis of particle swarm optimization algorithm, the thesis designs the WS small-world network with non-zero degree of node as a neighbor particle structure, adaptively adjusts population density, improves population diversity and introduces the boundary correction strategy to prevent the particles into the local optimal solution. An optimal model called Adaptive particle swarm optimization based neighbor (NAPSO). The time complexity, space complexity and the convergence of the algorithm were also analyzed. The Track of particles movement is stimulated in NetLogo, the phenomena of accumulation and spread shows NAPSO has better optimum capacity in later period of experiment.In the subjective evaluation, aiming at the consistency of the Analytic hierarchy process (AHP), it designs a NAPSO-CAHP that can modify the consistency of the matrix, which looks the function of coincidence indicator as the optimizing target. Moreover, it designs the Fuzzy AHP based on NAPSO, which uses the fuzzy complementary judgment matrix to replace the positive reciprocal judgment matrix, and bands the weight estimation and the testing of consistency of the judgment matrix together. In the objective evaluation, it designs the NAPSO-PP based on NAPSO banding the NAPSO and the projection pursuit model together, which looks the projection function value as the optimizing target. It computes the weights from the objective and the subjective evaluation, and then makes some combination and optimizing, get the differences among each indicator, so finally it can get the compounding weight. To sum up, it designs a NAPSO-PPAHP based on the external combination with PP and AHP, and a combined evaluation way of complex system based on NAPSOFAHP-PP.Evaluation examples showing calculation stability and high precision by using combined method of NAPSO and evaluation method of complex system, Compared with the existing methods, evaluation effect is better and classification has obvious effect. The result indicated that NAPSO is of certain significance in others of complex system evaluation and presents good effect on solving high-dimension and nonlinear problems.