Research On Models And Algorithms For Enterprises Project Investment Decision-making Under Uncertain Environments

With the intensification of market competition and the speeding up of economic globalization, enterprises are facing more and more uncertainties. How to make correct investment decisions and control risk is becoming an important issue that enterprises must face. Traditional investment decision theories often just consider the risks of investment from the static point of view. Therefore, enterprises using these models to guide investment decision-making practice can lead to make wrong decisions. Therefore, in order to describe accurate parameters of decision problems, it is necessary to set up uncertain programming model of investment decision-making problems, thus it can help decisions of enterprises more scientific. Although some scholars have used the stochastic programming model, but they mainly use the expected value model which is not suitable for the risk-averse decision makers. Multi-objective decision problems which can not be directly compared with each other haven’t been studied, the paper establishes stochastic goal programming model and fuzzy goal programming model based on satisfaction function. The existing algorithms for solving project investment decision model are very time consuming. The paper proposes that using the trained artificial neural network filters the population before Monte Carlo simulation. Only when individuals successfully passed the screening then the next step of the calculation will begin, so the computation time will be significantly less than previous methods. In order to solve risk management questions under uncertain environments, the paper constructs the sensitivity coefficients based on Spearman’s rank correlation coefficient and Monte Carlo simulation. The method is an important innovation in the risk management theory.The main innovations and contents that are researched in the paper are listed as follows:(1) How to build and solve model of the investment decision-making problems under random environments is studied in the paper. In random environments, the paper constructs distributed model, the expected value model and the chance constrained programming model by combining with the ideology of chance which was established by Charners and Cooper. The paper sets up the chance constrained programming model when the objective function also contains random variables and establishes the investment decision chance constrained model when enterprises use their own funds or use loans from financialinstitutions. When parameters of chance constrained stochastic obey normal distribution,the deterministic equivalent model is given. Because most of the time the chance constrained model of investment decision-making problem cannot be converted to deterministic equivalent model, the hybrid intelligent algorithm is proposed to solve the investment decision-making problem of chance constrained model based on Monte Carlo simulation and genetic algorithm. The paper uses numerical examples to illustrate the modeling idea of investment decision-making problem and the effectiveness of the algorithm.(2) The investment decision-making problem is studied when the fuzzy variable can be used to describe the investment decision-making model and solving methods are also discussed. The paper establishes two cases of models respectively when enterprises use its own funds or use loans and the deterministic equivalent model is given when the model of fuzzy parameters are normal distribution, triangular fuzzy variables. At the same time the paper gives a general hybrid intelligent algorithm for fuzzy programming model, a numerical example is given to illustrate the effectiveness of the modeling method and algorithm.(3) There are few scholars who study multi-factors sensitivity of investment decision problems under uncertain environments. The paper designs the method of calculating the sensitivity coefficient based on Spearman rank correlation coefficient and Monte Carlo simulation. The advantage of the method is that it is multi-factors sensitivity analysis.Because changes in one factor often accompanied by changes in other factors, multivariate sensitivity analysis considers this correlation, so it can reflect the combined effects of several risk factors and improve the limitations of single-factor analysis. Using this method it can be convenient to analysis the sensitivity of investment decision-making problem under uncertain environments.(4) General algorithm for solving investment decisions model under uncertain environments is slow and time-consuming, so the paper designs a new hybrid intelligent algorithm based on artificial neural network to accelerate the speed of solving the investment decision-making model under uncertain environments. The advantage of this method is that it reduces the time of Monte Carlo simulation which often is a time-consuming process, because the population were screened in advance by the trained artificial neural networks before Monte Carlo simulation. Only when individual successfully passed the screening then the next step of the calculation will begin, so thecomputation time will be significantly less than previous methods.(5) When the multiple targets of project investment decision problem cannot be directly compared with each other,stochastic goal programming model and fuzzy goal programming model based on satisfaction functions are established in the paper. The advantage of this method is that when the targets are converted to satisfaction, they can be directly compared with each other. For risk aversion type enterprises, they expect to maximize returns under a certain probability rather than directly to maximize expected return. The paper first time presents a new objective function for modeling investment decision problem by satisfaction function for each of company. Because satisfaction can be directly summed, thus solving not among the plurality of target problems.In short, the modeling methods for risk-averse enterprises and stochastic goal programming and fuzzy goal programming based on satisfaction function are discussed in the paper; Hybrid intelligent algorithms for solving all kinds of models and multi-factors sensitivity analysis for project investment decisions under uncertain environment are also in-depth studies. These methods will help enterprises making right project investment decisions under uncertainty environments. The numerical examples of the paper also confirm the validity of the decision-making method.