相关性分析-神经网络模型在宁夏用水量预测中的应用

Application of Correlation Analysis-Neural Network Model in Water Consumption Prediction in Ningxia

为了提高用水量预测精度,并且掌握各行业用水量比例。提出了相关性分析和多层感知器神经网络(MLP)耦合模型预测行业用水量,该模型首先利用相关性分析的方法筛选出对行业用水量影响较大的因子,再将主要因子数据输入到神经网络模型预测出行业用水量。进一步以地处干旱区域的宁夏回族自治区为例,提取2002—2016年主要影响行业用水的因子训练预测模型,以此模型预测2017—2020年的用水量并检验预测精度;预测结果显示,总用水预测值与实际值的多年相对误差均值仅为1.00%。最后使用该耦合模型对宁夏规划水平年2025年行业用水量进行预测,预测结果表明2025年宁夏总用水量有下降的趋势,这种变化趋势与自治区近几年大力推进节水型社会建设的政策相符合。

The research is conducted to improve the accuracy of water consumption prediction and grasp the proportion of water consumption in various industries.Therefore,a model coupling correlation analysis and multi-layer perceptron(MLP)neural networks is proposed to predict the water consumption of industries.In this model,correlation analysis is used to select factors that have a great impact on the water consumption of industries,and then the data of the main factors is input into the neural network model to predict the water consumption of industries.Taking the Ningxia Hui Autonomous Region in the arid region as an example,we extract the main factors affecting the water consumption of industries from 2002 to 2016 to train a prediction model and use this model to predict the water consumption from 2017 to 2020 for prediction accuracy verification.The prediction results reveal that the average value of the multi-year relative error between the predicted value of total water consumption and the actual value is only 0.93%.Finally,the coupling model is applied to predict the water consumption of industries in the target year of 2025 in the plan of Ningxia.The prediction results show that the total water consumption will decline in 2025,and this trend of change is consistent with the autonomous region's policy of vigorously promoting the construction of a water-saving society in recent years.