Construction and application of variable-weight buffer operators with perfect information
Zheng‐Xin Wang · Journal of Zhejiang University(Engineering Science) · 2013
To deal with the problem of the insufficient utilization of the behavior systems' information due to the simple structures of existing variable-weight buffer operators,a class of variable weights weakening and strengthening buffer operators with weights' regulating factors and perfect information were constructed based on the new information priority principle.Considering the nonlinear relationships between the weights' regulating factors and prediction errors,the genetic algorithm was employed to search the optimal regulating factor in interval(0,1),under the optimization objective of minimizing the average prediction error.The predition problems of the gross output of industrial enterprises in China and the foreign direct investment in Zhejiang Province of China were studied to verify the effectiveness of the variable-weight buffer operators with perfect information for dealing with the modeling problem of small sample disturbed data.The results show that the action process of the new buffer operators affecting original data embodies the new information priority principle.The optimization of the weight regulating factors can effectively control the interaction strength.The forecasting ability of the new buffer operators for disturbed systems is superior to the traditional high order buffer operators.