Study On Multi-objective Optimization Problem Of Investment And Returns Decision-making In Transport PPPs

Scientifically formulating an investment return strategy to achieve a reasonable sharing of the interests of all parties in Public Private Partnerships(PPP)implementation is the key to the success in transportation PPP project cooperation.According to the specific decision-making needs of each state in China's PPP implementation procedure,this dissertation aims to solve the multi-objective optimization problem(MOP)in the investment and returns decision-making of transportation PPP and then build up a whole decision-making method system for stages across PPP project approval.On the basis of problem description and theoretical analysis,sub-problems are defined as a transportation PPP return method classification problem,a general MOP,a Stackelberg decision MOP,and a sequential decision MOP.The main research works of this dissertation are concluded as follows:(1)A data-driven study on the classification of transportation PPP return pattern is carried out for providing a theoretical basis for transportation PPP to choose proper return pattern,including identifying significant influencing factors and generating intuitive classification criteria.Combining theoretical analysis and investment practice experience,the possible influence factors reflecting the project's own attributes are proposed,and new project data after the PPP trial-operation period are supplemented.Multi-nominal logistic regression(MNLR)model and Multi-ordinal logistic regression(MOLR)model are built to identify the key influencing factors,and C5.0 Decision Tree(DT)model is applied to mine the classification rules of return pattern.The differences in findings between this dissertation and existing studies are recognized as the local financial capability and the government debts are no longer the most important influencing factors in choosing transportation PPP return patterns;low marketability projects and county-level projects are likely to adopt the return patten of ‘fully paid by government'.(2)The general MOP of investment and returns strategy is studied to provide decision-making basis for determining the project transaction conditions and government guarantee conditions for the PPP implementation plan in ‘preparing' state.The method based on NSGAIII algorithm framework that integrating the integer truncation strategy is proposed to solve the built model.Based on that,current mature MOP algorithm NSGAII is used to solve the same problem,and the performances of the two algorithms are evaluated and compared through the convergence and diversity indicators.The results of a case study of the Beijing New Airport Line PPP project show that the given optimization scheme has improved the benefits of both parties when compared to the original scheme,and the solution method proposed can solve the Many Multi-objective problem(Ma OP)with mixed-integer variables,providing Pareto solutions in good convergence and diversity,which leave sufficient post-event decision space for the decision maker.Besides,the proposed the reference solution set construction strategy can overcomes the problem that real Pareto front is not available.(3)Under the circumstance that private sector and the government are in unequal decision-making status,the Stackelberg decision MOP is proposed to provide government decision-maker who adopt competitive procurement the optimized investment and returns strategy.According to the characteristic of Stackelberg decision,a Bi-level programming model is established.The upper problem is a MOP of cost paid by the government and users,while the lower problem is minimizing the net present value of the private sector's benefit with preset rate of return as the discount rate.A heuristic algorithm based on the NSGAIII framework is designed to solve the upper layer problem.Infeasible solution interception strategy and segmentation approximating optimization strategy of the lower problem are added to overcome the limitation that the traditional Bi-level programming algorithm requires the lower layer's reaction function is known to the upper layer.The validity of the model and algorithm is verified through the Big Outer Ring highway PPP project.It has found that when the total level of government payment in the project is the lowest,the actual returns level of private sector reached the lowest as well.Besides,a small amount external benefit backflow can achieve a significant reduction in the total amount of government subsidies.(4)In the PPP implementation stage,government will employ the dynamic revenue management on project.Considering the situation that the private sector deciding their level of effort according to the loss-and-benefit allocation decision,the problem of how to make the optimized allocation strategy under the triggered revenue management mechanism is proposed.Based on the Markov Decision Process(MDP)feature analysis of the proposed problem,a sequential decision MOP model is established.According to the characteristics of model-free and definite termination state of the problem,a strategy optimization method based on Monte Carlo sampling is designed,and the NSGAII algorithm solver is called to solve this problem.The effectiveness of the model and algorithm is verified by the above two PPP projects.Further,through the sensitivity analysis of the optimal strategy regarding the triggered allocation Threshold ROI changes,the problem of excessive rigidity of the existing revenue management measures is revealed.