基于非等时距加权灰色模型与神经网络的组合预测算法

Combination Forecasting Algorithm Based on Non-Equal Interval Weighted Grey Model and Neural Network

非等时距预测算法在不等时间间隔序列的趋势分析与预测方面具有重要作用.在传统灰色预测理论的基础上,提出一种基于非等时距加权灰色模型和神经网络的组合预测算法.通过构建非等时距加权灰色预测模型,将原始数据序列的平均值作为累加序列初值,将连续累积函数的积分面积作为背景值,对累加序列进行加权处理,以真实反映时间序列发展对预测结果的影响.在此基础上,引入BP神经网络对灰色预测的残差序列进行修正,进一步提高了预测精度.经算例验证,该算法预测精度达到1级,且高于类似算法.

The non-equal interval forecasting algorithm plays an important role in trend analysis and forecasting of sequences with different intervals. Based on the traditional grey forecasting theory, a combination forecasting algorithm based on non-equal interval weighted grey model and neural network was proposed. By constructing the non-equal interval weighted grey forecasting model, the average of original data sequence was regarded as the initial value of cumulative sequence, the integral area of continuous accumulation function was used as the back- ground value, and the cumulative sequence was processed by weighting in order to truly reflect the impact of time sequences development to forecasting results. On this basis, BP neural net- work was introduced to correct the residuals sequence of grey forecasting which further improved the forecasting accuracy. The numerical example indicates that the forecasting accuracy level of the algorithm is 1 and higher than similar algorithms.