Financing Risk Analysis Of Expressway Public Private Partnership Project Based On Support Vector Machine

For highway construction projects with large capital,difficult financing,long con:stru ction time and complicated engineering volume,solving the funding problem is the top p riority.To this end,C hina has actively introduced the PPP model,a public-private partnersh ip model prevalent in Europe,,which has greatly reduced the government’ s financial pres sure on related projects.However,there are advantages and disadvantages.The application t ime of PPP mode in China is still relatively short.Although the system has been establi shed,the development plan is not clear enough,and the related system guarantee is not p erfect.Therefore,there are still a lot of risks For PPP projects.The risk is related to the 1 evel of investment income,which is related to the success or failure of investment proje cts.This also explains the necessity of researching the financing risk of expressway proje cts under the PPP model.The main conclusions of this paper are:(1)Based on the research literature,this paper analyzes the financing risk of express way PPP project in detail according to the actual project.According to the risk performa nee form,the financing risk factors of expressway PPP project are divided into 7 categor ics,namely:Poliltical environmental risks,economic environmental risk,engineering construe tion risks,technical risks,operational management risks,environmental risks and force maje ure risks,and then the seven categories of risk factors are classified in detail,and a total of 25 risk indicators are classified to establish From the evaluation index system of hi ghway PPP project financing risk;(2)Combining with engineering practice,this paper introduces a machine learning m ethod-support vector machine regression(SVR)evaluation prediction mode to evaluate a nd predict the finaneing risk of highway PPP project,After example verification,support vector machine regression The forecasting model is indeed feasible and predictive in the financing risk assessment and predictiOon of the attack road PPP project,(3)On the basis of SVR,the heuristic intelligent algorithm genetic algorithm(GA)a nd ant colony algorithm(ACO)are used to optimize the support vector machine regress ion evaluation prediction model,and the mean square error MSE after parameter optimiz ation can be obtained.The MSE of the genetic algorithm optimization parameter is 0.001282,the MSE of the ant colony algorithm optimization parameter is 0.005684,and the M SE of the initial SVR is 0.01108.By comparing the mean square error MSE of the para meter optimization in the support vector machine model,the ant colony algorithm is Kon,wn.And the genetic algorithm does have an optimization effect in the parameter selectio n and optimization of the support vector machine mode,and the optimization effect of th e genetic algorithm on the parameters is relatively better.(4)By comparing the optimization effects of the two heuristic intelligent algorithms and the prediction stability of the optimized model,it can be seen that when the absolu te percentage error APE i.s used as the stability criterion,the prediction stability of the g enetic regression support vector machine(GA-SVR)is 1.6005%.The predictive stability of the ant colony regression support vector machine(ACO-SVR)is 5.3075%,while the i nitial stability of the initial regression support vector machine is 10.5275.1t can be seen that the optimization effect of the genetic algorithm is optimal and relatively stable.The applicability,accuracy and stability of the genetic regression support vector machine(G A-SVR)evaluation prediction model on the financing risk assessment and prediction of high,way PPP projects are proved;Expressway projects using the PPP model are basically projects with large capital a nd high technical requirements.From project establishment,construction,operation to hando ver,there are certain risk factors at each stage.The impact of these factors on the overall project is large.Or small.In the case of a large number of risk indicators and fewer tr aining samples,it is an efficient and accurate choice to use the small sample machine le arning method of support vector machine to evaluate and predict the risk.