Prediction Of Electronic Solid Waste Generation Based On Multivariate Grey Model

E-waste includes mobile phones,tablets,computers,televisions,refrigerators,air conditioners and other products that are common in daily life.E-waste has unique environmental and resource properties in solid waste.With dozens of recyclable materials such as copper,aluminum,iron,gold,silver,palladium and polymer,it is a large-scale "urban mine" with high utilization value.On the other hand,e-waste contains heavy metals and organic pollutants,which,if not disposed of properly,will have a serious negative impact on regional ecology and human health.Therefore,the recycling of electronic waste is of great significance to China’s environmental and economic benefits.In order to reasonably plan the recycling system and optimize the recycling process,timely and accurate prediction of the amount of e-waste production is particularly important.Multiple grey model is a common waste production prediction method,in the actual production life,due to the large number of random factors,time sequence,including electronic waste production,often showed obvious nonlinear and non-stationary,lead to simple use of multiple grey model to predict the results of the low precision,and can’t apply to different cities.Therefore,it is necessary to constantly improve and optimize the model to improve its prediction ability,versatility and intelligence.Combined with intelligent algorithm,this paper constructed a new model from the perspectives of hybrid modeling,parameter optimization of background value,correlation sequence coefficient and dynamic optimization of gray action,and applied it to the prediction of e-waste production volume and the prediction of social holding quantity of typical household appliances.The main research contents include:(1)In view of the nonlinear variation characteristics of the quantity of e-waste in the process of generation,an intelligent modeling method is proposed,which combines the multi-grey mechanism model with the neural network compensation model.In this method,BP neural network is used to compensate the error of grey mechanism modeling to improve the prediction accuracy of electronic waste production.Finally,the effectiveness of the proposed method is verified by using e-waste data from Washington State.(2)In view of the lack of direct data of e-waste production in China,and the reality of sufficient data of household electrical appliances,the forecasting method of household electrical appliances is proposed.In this method,grey correlation analysis was used to select the correlation factors of the population,and a hybrid model of multiple grey neural network was constructed.The intelligent algorithm was used to optimize the background parameters to improve the prediction accuracy of the model.Finally,the validity of the proposed method is verified by using the data of four typical household appliances in Jiangsu Province.(3)In view of the time-varying characteristics of the classical multivariate grey model’s correlation sequence coefficient and grey action,the exponential dynamic variable is taken as the correlation sequence coefficient and grey action,and the multivariate grey model optimized by power index is proposed,and the modeling method is systematically introduced.It is proved that the multivariate grey model optimized by power exponent has better predictive performance through comparative experiments.(4)Based on the proposed hybrid model of multiple grey neural network,the interface of e-waste prediction function module is designed for recycling enterprises to call.Using Django framework in Python language as the development tool,the e-waste prediction algorithm communicates with the sorting center system of recycling enterprises in the form of Web application program interface(Web API),and the e-waste prediction function module is designed to realize the application value of the proposed prediction algorithm.