河流水质时间序列分析的自组织预测方法及应用

应用前苏联学者伊万年科基于生物控制论中的自组织原理提出的一种数据组合处理方法，有效地解决复杂非线性系统预测。以汾河某段的水污染时间序列监测数据为基础，建立了一个河流水质污染预测的自组织模型，其建模样本的拟合值和检验样本的预测值相对误差分别在３．１％和５．３％以内。结果表明，自组织模型能较好地描述水污染时间序列数据之间的非线性关系，适合复杂水环境污染系统的预测工作。

A Self-organized Prediction Method of Time Series of Water Quality in River and its Applications

The self-organized prediction method was proposed by AG. Ivakhnenko, a Russian scholar, in the 1970's. It is a kind of non-linear group handling method of data and can solve effectively difficulty of data handling and fitting model of the non-linear complex system. In this paper, the self-organized method was applied to the time series analysis of water quality. The self-organized prediction model was built based on the field observation data of the time series of water quality, and the relative error of the predicted value of each fitting sample and testing sample was separately less than 3.1% and 5.3%. The results showed that the self-organized model could better describe non-linear relationship of the time series of water pollution and it is highly fit for the prediction analysis of complex water pollution system.