Research On Strategies Of Prediction And Maintaining Population Diversity For Multi-objective Evolutionary Optimization In Dynamic Environment

Evolutionary computation is an iterative search algorithm based on biological evolutionary mechanisms such as natural selection and genetic, and it is also an adaptive artificial intelligence technology based on Darwin’s theory of evolution by simulating the process of biological evolution and mechanism of self-organizing. Evolutionary algorithm has been has been successfully applied in the field of multi-objective optimization, but most of the research is mainly limited to the static multi-objective optimization problems. However, many real-world optimization problems are dynamic multi-objective optimization problems(DMOPs), with not only the conflict among multiple objectives but also the objective and related parameters may change over time. How to track the Pareto optimal solution set after the change is an important and challenging issue to solve the dynamic multi-objective optimization problems.In recent years, research on dynamic multi-objective evolutionary algorithm is still in its infancy. Researchers have designed many new ways on the basis of static algorithms to solve DMOPs, such as random initialization, hyper mutation, dynamic migration, memory and prediction et al. Those strategies have been proved by several researchers to be some effective methods to solve DMOPs. However, there are many defects in these methods along with the development of DMOPs, which mainly reflected in the following respects. Firstly, random initialization, hyper mutation, dynamic migration et al. strategies are all blind ways to enhance population diversity without a right guidance, the performance of convergence are unsatisfactory when dealing with more complex DMOPs. Secondly memory strategy reuses the optimal solutions which are previously searched by the memory to rapidly response to changes in the new environment. This strategy can achieve good results for periodic problems, but for non-periodic problems or in the first cycle of changing environment, population is still in the process of blind evolution, and algorithm is difficult to obtain a good convergence. Lastly, methods that based on prediction generate a new optimal solution set by the prediction model for the evolution of population, and help algorithm to respond quickly to new changes. So far, the accuracy of prediction is the main difficulty, how to design a more accurate prediction model is still the focus of the present research.In this paper, based on the researches and analysis of the current situation of domestic and foreign dynamic multi-objective evolutionary algorithm, a prediction strategy based on guide-individual and a novel prediction and memory strategies are proposed to solve DMOPs. In addition, a population diversity maintaining strategy based on dynamic environment evolutionary model is proposed.This paper proposes a prediction strategy based on guide-individual(GIPS). This strategy takes advantage of the evolutionary ability of the populations in the new environment, makes the population evolve independently after a short time and predict the direction of the optimal solutions. Compared with two state-of-the-art prediction-based dynamic multi-objective optimization algorithms, GIPS show faster response to the environmental changes.In this paper, we have proposed novel prediction and memory strategies(PMS). In the proposed prediction strategy, the combination of exploration and exploitation provides the basis for the algorithm to accurately predict the new optimal solution set. Meanwhile, the memory strategy allows the algorithm to better adapt to periodic problems, so that the algorithm can effectively solve the DMOPs with different variation. Compared with other three strategies, PMS has shown faster response to the environmental changes than peer strategies.Lastly, this paper from the dynamic environment of the train of thought, proposes a population diversity maintaining strategy based on dynamic environment evolutionary model(DEE-PDMS). This strategy builds a dynamic environment evolutionary model by grid, which makes use of the dynamic environment to record the different knowledge and information generated by population before and after environmental change, and in turn the knowledge and information guide the search in new environment. The model enhances population diversity by guided fashion, makes the simultaneous evolution of the environment and population.