Juan Zou

Novel prediction and memory strategies for dynamic multiobjective optimization

Dynamic multiobjective optimization problems (DMOPs) exist widely in real life, which requires the optimization algorithms to be able to track the Pareto optimal solution set after the change efficiently. In this paper, novel prediction and memory strategies (PMS) are proposed to solve DMOPs. Regarding prediction, the prediction strategy contains two parts, i.e., exploration and exploitation. Exploration can enhance the ability to search the entire solution space, making it adapt to the environmental change with a great extent. Exploitation can improve the accuracy of local search, making the algorithm to have a faster response to environmental change particularly in the solution set having relevance in the environment. In terms of memory, an optimal solution set preservation mechanism is employed, by reusing the previously found elite solutions, which improves the performance of the algorithm in solving periodic problems. Compared with two representative prediction strategies and a hybrid strategy combining prediction and memory both on seven traditional benchmark problems and on five newly appeared ones, PMS has been shown to have faster response to the environmental changes than the peer algorithms, performing well in terms of convergence and diversity.

The effect of diversity maintenance on prediction in dynamic multi-objective optimization

Abstract There are many dynamic multi-objective optimization problems (DMOPs) in real-life engineering applications whose objectives change over time. After an environmental change occurs, prediction strategies are commonly used in dynamic multi-objective optimization algorithms to find the new Pareto optimal set (POS). Being able to make more accurate prediction means the algorithm requires fewer computational resources to make the population approximate to the Pareto optimal front (POF). This paper proposes a hybrid diversity maintenance method to improve prediction accuracy. The method consists of three steps, which are implemented after an environmental change. The first step, based on the moving direction of the center points, uses the prediction to relocate a number of solutions close to the new Pareto front. On the basis of self-defined minimum and maximum points of the POS in this paper, the second step applies the gradual search to produce some well-distributed solutions in the decision space so as to compensate for the inaccuracy of the first step, simultaneously and further enhancing the convergence and diversity of the population. In the third step, some diverse individuals are randomly generated within the region of next probable POS, which prompts the diversity of the population. Eventually the prediction becomes more accurate as the solutions with good convergence and diversity are selected after the non-dominated sort [1] on the combined solutions generated by the three steps. Compared with three other prediction methods on a series of test instances, our method is very competitive in convergence and diversity as well as the speed at which it responds to environmental changes.

A population diversity maintaining strategy based on dynamic environment evolutionary model for dynamic multiobjective optimization

Maintaining population diversity is a crucial issue for the performance of dynamic multiobjective optimization algorithms. However traditional dynamic multiobjective evolutionary algorithms usually imitate the biological evolution of their own, maintain population diversity through different strategies and make the population be able to track the Pareto optimal solution set after the change efficiently. Nevertheless, these algorithms neglect the role of dynamic environment in evolution, lead to the lacking of active and instructional search. In this paper, a population diversity maintaining strategy based on dynamic environment evolutionary model is proposed (DEE-PDMS). This strategy builds a dynamic environment evolutionary model when a change is detected, 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. A comparison study with other two state-of-the-art strategies on five test problems with linear or nonlinear correlation between design variables has shown the effectiveness of the DEE-PDMS for dealing with dynamic environments.

On decomposition methods in interactive user-preference based optimization

Graphical abstractDisplay Omitted HighlightsA model how to combine preference information and the framework of MOEA/D.A new way of generating weight vectors with preference information.Use the ASF function into MOEA/D to reflect preference information.The proposed approaches could deal with many-objective problems well.The desired solutions can be found in different ROIs interactively. Evolutionary multi-objective optimization (EMO) methodologies have been widely applied to find a well-distributed trade-off solutions approximating to the Pareto-optimal front in the past decades. However, integrating the user-preference into the optimization to find the region of interest (ROI) [1] or preferred Pareto-optimal solutions could be more efficient and effective for the decision maker (DM) straightforwardly. In this paper, we propose several methods by combining preference-based strategy (like the reference points) with the decomposition-based multi-objective evolutionary algorithm (MOEA/D) [2], and demonstrate how preferred sets or ROIs near the different reference points specified by the DM can be found simultaneously and interactively. The study is based on the experiments conducted on a set of test problems with objectives ranging from two to fifteen objectives. Experiments have proved that the proposed approaches are more efficient and effective especially on many-objective problems to provide a set of solutions to the DMs preference, so that a better and a more reliable decision can be made.

A prediction strategy based on center points and knee points for evolutionary dynamic multi-objective optimization

Abstract In real life, there are many dynamic multi-objective optimization problems which vary over time, requiring an optimization algorithm to track the movement of the Pareto front (Pareto set) with time. In this paper, we propose a novel prediction strategy based on center points and knee points (CKPS) consisting of three mechanisms. First, a method of predicting the non-dominated set based on the forward-looking center points is proposed. Second, the knee point set is introduced to the predicted population to predict accurately the location and distribution of the Pareto front after an environmental change. Finally, an adaptive diversity maintenance strategy is proposed, which can generate some random individuals of the corresponding number according to the degree of difficulty of the problem to maintain the diversity of the population. The proposed strategy is compared with four other state-of-the-art strategies. The experimental results show that CKPS is effective for evolutionary dynamic multi-objective optimization.

Adaptive neighborhood selection for many-objective optimization problems

Abstract It is generally accepted that conflicts between convergence and distribution deteriorate with an increase in the number of objectives. Furthermore, Pareto dominance loses its effectiveness in many-objectives optimization problems (MaOPs), which have more than three objectives. Therefore, a more valid selection method is needed to balance convergence and distribution. This paper presents a many-objective evolutionary algorithm, called Adaptive Neighborhood Selection for Many-objective evolutionary algorithm(ANS-MOEA), to deal with MaOPs. This method defines the performance of each individual by two types of information, convergence information (CI) and distribution information (DI). In the critical layer, a well-converged individual is selected first from the population, and its neighbors, calculated by DI, are pushed into neighbor collection (NC) soon afterwards. Then, the proper distribution of the population is ensured by competition individuals with large DI go back to the population and individuals with small DI remain in the collection. Four state-of-the-art MaOEAs are selected as the competitive algorithms to validate ANS-MOEA. The experimental results show that ANS-MOEA can solve a MaOP and generate a set of remarkable solutions to balance convergence and distribution.

Many-objective optimization based on information separation and neighbor punishment selection

Many-objective optimization refers to optimizing a multi-objective optimization problem (MOP) where the number of objectives is more than 3. Most classical evolutionary multi-objective optimization (EMO) algorithms use the Pareto dominance relation to guide the search, which usually perform poorly in the many-objective optimization scenario. This paper proposes an EMO algorithm based on information separation and neighbor punishment selection (ISNPS) to deal with many-objective optimization problems. ISNPS separates individual’s behavior in the population into convergence information and distribution information by rotating the original coordinates in the objective space. Specifically, the proposed algorithm employs one coordinate to reflect individual’s convergence and the remaining

A dynamic multiobjective evolutionary algorithm based on a dynamic evolutionary environment model

m-1