Abstract:

The support vector machine (SVM) algorithm is of reliable global optimality and good generalization，suitable for the learning of finite samples. However，the results considerably depend on the SVM model parameters and the conventional parameter choosing method by experience is unsatisfactory. Using the particle swarm optimization (PSO) random search strategy, we can establish the optimization parameters of support vector machine. It is shown that ACOSVM is much better in the simulation results than the artificial neural network，which greatly improves in fitting precision，and it has good generalization ability.

Key words: support vector machine;particle swarm optimization;parameter optimization;water warehouse water level