基于决策树技术及在线监测的水质预测

利用北方某城市水源的水质在线监测系统,建立了基于决策树技术,具有较强可视性和实际应用,以及能预测次日源水中叶绿素水平的决策树模型.该模型将某城市水源在线监测的溶解氧和太阳辐射照度数据转换计算为每日平均标准偏差及均值,并与每日定时取样测定的叶绿素含量一起作为预测因子,通过将115组数据的前100组数据作为训练集建立预测次日叶绿素水平决策树模型,并采用后15组数据进行模型的仿真预测检验,结果只有3 d的预测出错,预测准确率达80%.并讨论了模型建立对数据的要求及解读预测规则等问题.

ForecastWater Quality Based on Decision making TreeandOnlineMonitoring

This article set up the technology of decision-making tree used the on-line water monitoring system from the water source of a North-China city, and it has strong objective and practical application, as well as the new decision-making model can predict the chlorophyll level of water sources next day. The model convert the dissolved oxygen from online monitoring in urban water source and the data of solar radiation illumination into the average daily standard deviation and dispersion, which as the forecast factors with the regular daily sample measurement of chlorophyll content. There are 115 sets of data, the previous 100 sets of them will be as a training group to establish the decision-making tree model for forecasting the chlorophyll levels of the following day,the follow 15 sets of them used as forecasting test results of simulation models. The result has shown only 3 d forecast was error, the accurately rate is 80%. In addition it discussed the data of model building and interpretation of forecasting rules.