The GM(1,1) Model Based Particle Swarm Optimization And Its Application

The management of reservoir is a complex system. According to the analysis of mini-type water and electricity system in SanYa-city, a project which based on the power load forecasting is made to manage reservoir. And then, we develop a suit of user software of management of the mini-type water and electricity system. We propose a series of measures to optimize the basal grey forecasting model when we used it to predict some data: If we add a constant number a0 to each number of the original series and build grey system model with the new series, a0 will affect the precision of the model. The best a0 value can be found out after analyzing the error of the different offset. We can get a new series with substituting the new data for the oldest data ,and without changing the dimension of the original series. We call the new series invariable-dimension new-informational series. The model which based on it called invariable-dimension new-informational model. It can improve the precision and can be used repeatedly. Remained error model is based the remained error series from the basal model. It is used to modify the basal model and improve the precision. The parameter αin background of GM(1,1) z ¹ (k) = αx ¹ (k -1) + (1 -α)x ¹(k) will influence the precision too. Particle Swarm Optimization is better than other method in choosing the best a0 and parameter α. We get a useful optimizing model from synthesizing the above method. It is well done in the example. In fact, the software based on this optimizing model is of great benefit to the management of the reservoir system.