Selective ensemble simulate meta-model based-on global optimize strategy

The increment of model complexity and size has been bottle neck of improve simulation system analyze emulation effective and decision maker cognize complex system. One of the effective methods to solve this problem is to replace complex physical model with simple simulate meta-model. Aim at slowly modeling speed and difficulty to effective updating problem using traditional neural network and other machine learning based simulate meta-model algorithm, and lower modeling accurate and generalization et al problems, a new global optimization based selective ensemble strategy is proposed in this paper, and single-hidden layer feed-forward networks with random weights (SLFNrw) is used to construct simulate meta-model. At first, simulate meta-modeling technology using in complex system simulation is analyzed. Then, global optimization based selective ensemble SLFNrw simulate meta-modeling strategy and algorithm are clarified in detail. At last, synthetic function and benchmark data are used to test the proposed algorithm. The results show the proposed algorithm can obtain well trade-off between modeling accuracy and speed, which can be widely used in complex system analysis based on simulation.