Research On Coevolutionary Algorithms And Its Application In Portfolio Investment

Evolutionary algorithm(EA) is a kind of stochastic optimization methods, EA with the features of self-organization and self-adaptive is established by simulating a natural evolutionary process. Coevolutionary algorithm(CEA) is a novel evolutionary algorithm inspired by the principle of biological evolution. It is believed that there is a certain relationship between different populations, e.g., competition or cooperation. Populations, strategies or operators cooperate in CEA in order to enhance the CEA's adaptation in more complex dynamic environment. Compared with conventional evolutionary algorithms, CEA not only can achieve a better performance, but also have advantages of stability and robustness. According to the different methods of algorithm design, CEA can use a muti-population pattern or a mixed strategy pattern. In the thesis, two CEAs based on mixed strategy are proposed and applied in the fields of single-objective optimization and multi-objective optimization. The main contributions are detailed below:(l)Based on some research on related algorithms which have its advantages and disadvantages by using a pure strategy, a mixed strategy CEA using operators of single-point and particle swarm(MSPSO) is proposed. Simulation results show that the mixed strategy can overcome the shortcomings of a pure strategy and MSPSO is superior to several popular algorithms proposed in recent years.(2)A new CEA to solve multiobjective optimization problems is presented. The new approach uses single-point operator to improve the performance of CMOPSO algorithm. After a set of test problems, results shows that the new approach can not only ensure its performance of convergence and distribution, but also have the advantages of stability and robustness.(3)The application of CEA is studied. For the portfolio investment problem, this thesis proposes to use the new mutiobjective CEA approach which is not sensitive to model. The new approach can get the pareto front of portfolio investment problem directly and it is a convenient and efficient way.