Simulation And Prediction Of The Amount Of The Scraped Car Based On Fuzzy Colored Petri Net In Reverse Logistics

The continuing growth of car ownership make the rising number of end-of-life vehicles, which makesthe reasonable and effective treatment of the solid waste produced by end-of-life vehicles particularlyimportant.The effective prediction of end-of-life vehicles can not only make the resources of end-of-lifevehicles rational utilization,but also have a guiding role on the decision of policies and regulations.Incurrently,the prediction methods for automobile production and sales emerge in endlessly,but little for theforecast of end-of-life vehicles.Formerly,the number of end-of-life vehicles are predicted by5%~8%ofcar ownership. Because the continuous improvement of automotive technology and the road conditions,which make the pace of end-of-life vehicles always lag behind the pace of automobile production andsales.It is visible and inappropriate that the prediction of end-of-life vehicles is achieved by a singleindicator about car ownership.This method can’t take into full account the life circle factors of vehicles andexternal economic factors’s influence on the amount of end-of-life vehicles. Therefore,aiming at the aboveproblem,this paper proposes the complete logic method.It is fuzzy colored petri nets.The method combinesfuzzy reasoning with colored petri net system theory and establishes the simulation prediction model ofend-of-life vehicles in various areas based on fuzzy colored petri net (FCPN).Firstly, this paper analyses the influence factors of end-of-life vehicles from the life circle factors ofvehicles and external economic factors two aspects. After weighing the ease or complexity of indexquantification and the integrity of the collected datas,we screen out12indexes as the characteristic indexesof end-of-life vehicles.The indexes include such as GDP, per capita disposable income, population density,new registered civil cars, car ownership, car production and so on.The corresponding datas are collectedand the determined indexes are achieved by quantization.Because the dimension difference exists amongvarious indicators datas,the standardization of indexes are achieved.The more indicators could make therelationship structure of model extremely complex and too much indicator variables will greatly increasethe number of iterations of model.In order to simplify the model structure, the method of principalcomponent analysis is adopted to analyse12indicators variables. We extract the first four principalcomponents as the input of the FCPN through the size of the accumulated principal components variancecontribution rate.The degree of interpretation of the extracted principal components are performed to bevalidated.At finally,we establish the prediction model of end-of-life vehicles in various areas based on FCPN and the model is adopted to predict the number of end-of-life vehicles in various areas effectively.Onthe basis of the result, the standard error (MSE) and root mean square error (RMSE), mean absolutepercentage error (MAPE) and mean absolute error (MAD) are as the evaluation indexes.The papercombines with the algorithm instance and compares FCPN prediction model with multivariate regressionmodel and RBF neural network model. The analysis results show that the constructed FCPN models havethe highest accuracy.