Helper Objective Strategy-Based Intelligent Optimization:Methods And Application

Complex optimization problems arise regularly in the fields of national defense,industry,agriculture,transportation,management and so on.These problems show the characteristics of high-dimensional,nonlinear,multi constraints,mixed discrete and continuous variables,which brings great challenges to the traditional optimization methods.In recent years,intelligent optimization algorithms based on population iterative search have become an important method to solve nonlinear and complex optimization problems because of simplicity and efficiency,especially low requirement for problem structure information.However,the current research on efficient solutions to complex optimization problems mostly focuses on the innovation or improvement of the algorithm itself,such as designing new population initialization method,new individual crossover,mutation strategy,new individual sorting mechanism,etc.few scholars pay attention to how to gradually solve the original optimization problem from the perspective of the problem by constructing optimization tasks similar to the original optimization problem,i.e.helper tasks.Therefore,this study investigates the intelligent optimization method based on helper objective technology,and puts forward a new framework to solve complex optimization problems.The core idea is as follows: for the general intelligent optimization algorithm is easy to fall into local optimization,a helper optimization objective function similar to the original optimization objective function but easier to solve is proposed.For example,for multi-modal optimization problems,a single peak helper objective function close to the global optimal solution is constructed,and the useful knowledge extracted in the process of solving the helper objective function is used to enhance the ability of the algorithm to jump out of local optimization,so as to effectively solve the original optimization objective function.In detail,the helper objective function design methods for discrete and continuous optimization problems are studied in this paper.For discrete optimization problems,two methods of designing helper objectives are proposed: one is to simplify the original objective function;another is to simplify decision variables.Aiming at the continuous optimization problem,this paper proposes a helper objective design method based on agent model.The commonly used agent models for optimization include Gaussian process,polynomial regression,radial basis function,etc.Especially,Gaussian function has significant global advantages and smooth characteristics.Therefore,Gaussian process fitting function is selected to construct helper objectives.In order to make effective use of the helper objective function,the original objective and helper objective are regarded as two related tasks through the multi-task optimization framework.Using the idea of knowledge transfer,the useful information obtained from solving the helper objective task is transferred to the task of the original optimization objective function in the process of algorithm search iteration,so as to speed up the optimization of the original objective task.Finally,the effectiveness of the helper objective strategy is verified by an international standard example.In order to further verify the practicability of the algorithm,the paper carried out the application research of multi-UAVs task planning based on helper objective technology,and established the multi-UAVs task allocation and path planning model in the battlefield individual soldier energy support scenario.For the asymmetric discrete task allocation optimization,the helper objective function is constructed by linearizing the original objective function,while for the continuous path planning considering spatial cooperation,the helper objective function was constructed based on Gaussian process,The simulation results show that the proposed intelligent optimization method based on helper objective technology can effectively solve the problems of multi UAV task allocation and path planning,and has good application value.