Analysis Of Residential Water Consumption And Web System Design Based On Multi-model Predictio

With the rapid development of the economic level and the increase of the urban population,the water consumption of urban residents has increased sharply.Against the backdrop of water shortage in China,the imbalance between supply and demand of urban water resources is becoming increasingly serious.As the terminal link of urban water supply,accurately predicting household water consumption is conducive to reasonable scheduling of urban water supply systems,and can alleviate the contradiction between supply and demand of urban water resources to a certain extent.Based on previous studies,this paper aims at the respective characteristics of ARMA model,metabolic GM(1,1)grey prediction model,and BP neural network model.An optimized combination forecasting method based on ARMA-BP and grey neural network is proposed to predict household water consumption.As the preliminary work of the model prediction process,this paper takes the daily water consumption of a residential area in Rizhao City as the original data.Firstly,this paper analyzes the sequence characteristics of the dataset and then performs missing value supplementation and outlier elimination on the original data.It also analyzes the influencing factors of residential water consumption and uses grey correlation analysis and Pearson correlation coefficient method to determine key influencing parameters of residential water consumption,including historical water consumption data,daily average temperature,whether it is a holiday,stepped water price,sunshine duration,and precipitation.Finally,this paper uses ARMA model,metabolic GM(1,1)model,BP neural network model,ARMA-BP combined forecasting model,and gray neural network combined forecasting model to predict daily water consumption of residents.The prediction results show that the ARMA model has a good prediction effect on the linear characteristics of water consumption prediction,but it is difficult to capture the non-linear characteristics of the series.The average relative error of the model is 7.93%,and the accuracy of the model is relatively low.Compared with the ARMA model,the GM(1,1)grey prediction model of metabolism can achieve better approximation of nonlinear characteristics,and the model accuracy has also been greatly improved.The average relative error of the model has reached6.17%.Compared with the previous two models,the BP neural network prediction model has improved both in terms of nonlinear characteristics and model accuracy to a certain extent,with an average relative error of 4.65%.By comprehensively comparing the average relative error of the three prediction models,the prediction results are consistent with the trend of water consumption,but the accuracy is relatively low.Combining the advantages of different algorithms,this paper proposes a parallel combination method to establish an ARMA-BP and gray neural network optimization combination prediction model.Compared with a single combination model,the optimized combination model has higher prediction accuracy.The average relative error of ARMA-BP and grey neural network combination model are 3.29% and 3.52%,respectively.On the basis of water consumption prediction,in order to enable residential users to better grasp the status of household water consumption,this study uses a front and rear end separation architecture to design a household water consumption prediction system for residents.The backend of the paper uses Java language for development and the front-end uses HTML+CSS+JS for webpage production.In terms of functional structure,the system separates the Web side from the algorithm side.It can not only improve the utilization rate of the algorithm,but also update or replace the algorithm according to the actual situation.In addition to the water consumption prediction function,this system can also achieve household water consumption analysis,user information management,and other functions.Users can view the usage of household water consumption through the visual interface of the Web application system,and can also predict future water consumption data and reasonably adjust water use plans,indirectly improving people’s awareness of water conservation and contributing to urban water conservation work.