Prediction of water consumption in Baotou based on amended combination model
[Objective] This study established a combination model of Grey model and BP neural network revised by Markov chain to better forecast water consumption.[Method] Main factors influencing change of water consumption in Baotou were analyzed by gray correlation analysis,and a combination model of Grey model and BP neural network revised by Markov chain was established.The model was used to forecast water consumption in 2009 and 2010 in Baotou,and the model prediction was compared with results of Grey model,BP neural network and combined Grey neural network.[Result] Water consumption in Baotou was influenced by population,GDP,total industrial output value,built-up area greenbelt cover rate,cultivated area,and industrial water recycling rate.Comparison of water consumption in 2009 and 2010 of Baotou predicted by the established combination model of Grey model and BP neural network revised by Markov chain and actual water consumption indicated that the relative errors were 0.16％ and 2.16％ respectively,and the relative error of root mean square was 1.53％.The relative errors of root mean square for Grey model,BP neural network and combined Grey neural network were 4.34％,3.08％ and 1.99％,respectively.The combination model of Grey and BP neural network revised by Markov chain was the best.[Conclusion] Combined Grey-neural network model revised by Markov chain had high prediction accuracy.
Baotou citywater consumption predictioncombination model of Grey model and BP neural network modelMarkov chain
NETL
NSTL
万方数据
维普
