Memetic-based Multi-objective Evolutionary Optimization Algorithms And Application

Interval multi-objective optimization problems(IMOPs)are very ubiquitous and important in real production and life.Because of its objective function with uncertain parameters,it is very difficult to use existing multi-objective optimization evolutionary algorithms(MOEAs)to solve this kind of problems.At present,IMOPs is one of the focuses of evolutionary optimization,and memetic algorithm(MA)is an effective method to deal with this kind of optimization problem.In addition,this paper use surrogate model to simplify the local process and can ensure the accuracy of the results under the premise of greatly improving the efficiency of the operation.Based on this premise,this paper presents a multi-objective evolutionary optimization algorithm for interval parameter based on surrogate model,which will be described in detail below.Firstly,this paper proposes an interval multi-objective evolutionary optimization algorithm based on memetic algorithm(IMOMA),which integrates an improved local search.This algorithm mainly includes two parts: global search and local search.The global search adopts the existing IP-MOEA based on interval dominance.In the local search process,there are three key techniques,namely,local search activation mechanism,local search initial population establishment and local search strategy.By using 10 benchmark IMOPs and an uncertain optimization problem in solar desalination,IMOMA can obtain the approximate Pareto optimal solution set with good convergence,good distribution and little uncertainty,and compared with the IP-MOEA without local search.The final result shows that the proposed IMOMA is superior to IP-MOEA.However,for the hypervolume,the time complexity of IMOMA is too large and the operation efficiency of this method is low.Then,for the low efficiency of IMOMA,the surrogate model is incorporated into the complex problem of local search,and a multi-objective optimization algorithm of memetic interval based on this local surrogate model(SS-IMOMA)is proposed.The overall architecture still continues IMOMA design,the difference is mainly reflected in the following aspects: after obtaining the local initial population,this population will be divided into the train set and test set,which is used to train the surrogate model and verify the accuracy of this model,repsectively.In the subsequent local search strategy,the improved single-objective fitness function is used to evaluate the contribution to hypervolume and uncertainty of the individual,and the support vector machine(SVM)is used to structure the single-objective fitness evaluation to achieve the purpose of improving operational efficiency.In the same way,SS-IMOMA has better algorithm's performance than IP-MOEA without local search for 10 benchmark IMOPs and an uncertain optimization problem in solar desalination,and the time complexity of SS-IMOMA is smaller than the IMOMA without surrogate model.Finally,the above method is applied to the solar desalination problem,and uses MATLAB to provide a GUI environment to design an optimization platform.The output of the original data is interval-processed to solve the problem of uncertainty in the actual engineering.Then,this paper analyzes the data of the interval parameters with support vector machine regression analysis,and constructs the mapping relationship between input and output by support vector machine,which can solve the difficulty of numerical model establishment.Based on the interval dominance relation,the optimal solution set can be obtained by using SS-IMOMA for the purpose of obtaining the optimum working condition of solar desalination system for hot air temperature and other variables.In order to modify the parameters easily,intuitive display of the results,the system provides optimization parameters,regression curves and other images of the display module,theand can visually display the output.The platform provides the results of parameter optimization in the regression process,the results of the regression curve and other images of the display module,the optimal solution set of the output module,the algorithm's parameters settings module,and open files and other controls design.The proposed IMOMA can provide a reliable and effective solution for multi-objective optimization of interval parameters.In view of the low efficiency of IMOMA,this paper proposes SS-IMOMA solve this problem successfully.In particular,IMOMA and SS-IMOMA can obtain a better approximate Pareto optimal solution set for the uncertain benchmark problem and the uncertain optimization problem in solar desalination.