Font Size: a A A

The Research Of Intelligent Optimization Algorithm And Their Applications To Electromagnetic Inverse Scattering

Posted on:2016-08-20Degree:DoctorType:Dissertation
Country:ChinaCandidate:J LuFull Text:PDF
GTID:1108330479499354Subject:Electrical engineering
Abstract/Summary:PDF Full Text Request
The main task of electromagnetic inverse scattering is to rebuild on the target by measured scattering field data. Because linear methods are more restricted, in this article, the inverse scattering problem was transformed into an optimization problem firstly, and then was dealt with optimization algorithm. Several classical intelligent optimization algorithms have problems such as premature convergence, low efficiency of global search etc., therefore, exploring a new optimization algorithm is imminent. Harmony Search algorithm is a new intelligent optimization algorithm. It has been successfully applied to many fields and showed superior performance. Based on the thorough analysis of intelligent optimization algorithm and electromagnetic inverse scattering theory, Harmony Search algorithm basing on multimode optimization technique and its optimization on electromagnetic inverse scattering problems were studied, the main work and innovation include the following aspects:1. According to the replacement error of standard crowding strategy, introduced crowding distance which could limit replacing error in a certain range, Limiting Crowding Harmony Search algorithm was proposed. According to the poor niche maintaining ability of Limiting Crowding strategy, introduced dynamic crowding factor, put forward Dynamic Crowding strategy. Experiments showed that the crowing strategy could effectively increase the diversity of individuals, and avoid trapping in a single optimal solution.2. Topology structure of the objective function in electromagnetic inverse scattering is too complex to make peak radius available. Aiming at that defect, a method of automatic adjustment of peak radius was studied, sharing strategy of Adaptive Adjusting Peak Radius was proposed. Because computation of individual’s sharing fitness in sharing policy was time-consuming, Clearing Sharing Harmony Search algorithm was proposed. Experimental results showed that the two share strategies had good optimization effect, and Clearing Sharing policy could effectively reduce time complexity of algorithm.3. Traditional method of population division is mechanical, in order to solve this problem, studied the method of adjusting peak radius automatically based on the characteristics of problem, made each child population cover a single peak only, put forward Topology Multi-population policy. Experimental results showed that its optimization effect was very ideal for most test functions.4. For multimode optimization need to protect the diversity of individuals in harmony memory matrix, multi-objective optimization technique was introduced into the multimode optimization problem, increased measurement of crowding distance between individuals as the second goal, the original single objective optimization problem was converted into a bi-objective optimization problem. On the basis of NSGA-II, Non Dominated Sorting Harmony Search algorithm was proposed. Experimental results showed that, for most test function optimization, its effect was perfect, and it could achieve convergence in less number of iterations.Finally, these harmony search algorithms based on multimode optimization strategy were applied to some electromagnetic inverse scattering problems, and obtained satisfactory results, full experimental data confirmed correctness, validity and superiority of the algorithms.
Keywords/Search Tags:harmony search, multimodal optimization, bi-objective, sharing, multipopulation
PDF Full Text Request
Related items