天然气需求自适应优化组合预测模型研究

STUDY ON PREDICTION MODEL OF GAS DE-MAND SELF-ADAPTING OPTIMIZATION COM-BINATION

文章用线性回归预测法、人工神经网络预测法、灰色系统预测法等对天然气消费需求量进行预测后,发现各预测值与实际需求量之间存在较大误差。为综合利用各方法提供的信息,避免单一预测模型丢失有用信息,减少随机性,提高预测的准确性,采用了新的优化组合预测法对天然气需求量进行预测,预测结果比单一预测法好。但优化组合预测模型求解的权系数是固定的,对有众多影响因素的天然气需求量进行预测不是最适合的,为此采用自适应递推适时算法对其进行改进,使用变化的权系数进行预测。结果表明自适应递推优化组合预测模型比优化组合预测模型预测的结果还要好。

Several predictions for gas demand are conducted with the linear regression method, the artificial neural network method, the gray system method, etc. The results show there are big errors between the predictions and the real demands. To use the data obtained from the different methods comprehensively, prevent the defects for a single method to loss some useful data, reduce the randomness, and improve the prediction accuracy, the prediction method with optimization combination is applied for gas demand prediction. The predicted result is better than that with a single method. But the weight factors of the solution for the prediction model with optimization combination are constants, which are not the fittest for the prediction of gas demand with multiple influence factors. So, the self-adaptive timing arithmetic is used to improve the model. The results show the prediction model with self-adaptive timing optimization combination is even better than that with optimization combination.