Abstract:Grey prediction algorithms are studied in this paper. In GM (1,1) model, development coefficient a and grey action quantity u are two key parameters which have a great impact on prediction system. In traditional methods, these two parameters are obtained by a least squares method with high computation overhead and large prediction error. This problem is discussed in this paper and a grey prediction method called PSOGP is proposed. Based on the GM (1, 1) model, PSOGP uses a particle swarm optimization algorithm to solve the two parameters. Simulation results show that, comparing with the classic GM (1,1) model, the accuracy of PSOGP is greatly improved.