Dynamic multi-objective evolutionary method based on transfer learning and special point strategy

Abstract

The embodiment of the invention provides a dynamic multi-objective evolutionary method based on transfer learning and a special point strategy, and relates to the field of evolutionary computation. The method comprises the following steps: analyzing a dynamic multi-objective optimization scene to obtain an initial population of the optimization scene; optimizing the initial population by adoptinga multi-objective estimation distribution method, and obtaining a front-end optimization set of the initial population; acquiring special points according to the front-end optimization set; when the decision variable changes, mapping the special point to a high-dimensional space based on a joint distribution adaptation transfer learning method to obtain the special point at the next moment; and optimizing the initial population at the next moment by adopting a multi-objective estimation distribution method, and obtaining a front-end optimization set of the initial population at the next moment. Through the special points, accurate prediction of the special points in the new environment is achieved, other individuals are generated randomly, the population diversity is increased, and therefore the convergence speed and the convergence precision of the algorithm in the new environment are improved.

Description

technical field [0001] The invention relates to the field of evolutionary computing, in particular to a method for solving a monthly centralized bidding mechanism based on a co-evolutionary algorithm. Background technique [0002] In many optimization fields such as production scheduling, artificial intelligence, combinatorial optimization, engineering design, large-scale data processing, urban transportation, reservoir management, network communication, data mining, and capital budgeting, many complex problems that are closer to real life are often encountered. Dynamic and static optimization problems. In the past few decades, people mostly devoted themselves to the research of static target problem, and until recent years, the dynamic target problem has attracted more and more researchers' interest. [0003] At present, there are not many research results on dynamic multi-objective problems, and it has just started in the world, and there are few theories that can be seen...

AI Technical Summary

Problems solved by technology

Although the above-mentioned various implicit memory methods can enable the evolutionary algorithm to indirectly store some effective information, it is not sure whether the algorithm can effectively use this information

3) Prediction-based methods: For example, IasonH et al. proposed a forward estimation method (Forward-Looking Approach) based on solving the dynamic single-objective optimization evolutionary algorithm, and Aimin Zhou et al. proposed a dynamic multi-objective evolution based on population prediction Algorithms, ArrchanaMuruganantham et al. proposed an algorithm based on Kalman filter prediction, and Juan Zou et al. proposed methods based on central point and inflection point prediction strategies that can respond quickly after environmental changes, but the accuracy of prediction is the main difficulty

[0004] At present, most dynamic multi-objective optimization algorithms predict that the solution sets in different environments follow independent and identical distribution. This assumption undoubtedly simplifies the complexity of the problem, but changes in the frontier may lead to different distributions of solution sets in different environments. This is very difficult to solve for traditional machine learning. Min Jiang et al. proposed a dynamic multi-objective optimization algorithm based on transfer learning, using the transfer component analysis (TCA) transfer learning method to map the current frontier to a high-dimensional space, and obtain The initial population at the next moment, this method effectively improves the quality of the obtained solution, but the calculation is very complicated

Images