改进GM(1,1)模型在环境噪声预测中的应用

Application of Improved GM(1,1) Model for Noise Prediction

从传统GM(1,1)预测模型构造原理出发分析其存在的理论缺陷,此模型适用于具有较强指数规律的序列,只能描述单调变化过程,而且预测数据序列中出现连续两个相等值时将使得a等于0,会使预测数列发散致使预测失效。采用平均斜率的概念计算α(k)序列,并利用MATLAB编程实现数据处理。广州市环境噪声污染数据计算表明,新方法有满意的拟合和预测效果,从而为提高建模精度提供了新的途径。

In this paper the structure principle of GM （1, 1） was analyzed and the defect was pointed out that the traditional model only adapt to the data with exponential law, and only can fit for the monotonic data. The prediction is defected when there are two consecutive data unchanged, and ‘ α＇ is equal to 0.5 in prediction formula. The average slope is used to calculate a （ k ） instead of constant 0. 5, and program MATLAB is used to deal with the tested data. An actual example of Guangzhou city showed that the new model has satisfactory fitting effects and thus points a novel direction to higher modeling procedure.