The Improved Technologies Of Gravity Search Algorithm And Their Applications To Antenna Array Syntheses

Array synthesis is one of the core problems in the design of array antenna.The efficient array synthesis algorithm is helpful to design the desired array antenna with high performance.As a heuristic algorithm,gravity search algorithm has strong computational power,wide adaptability and strong global convergence.It has been successfully applied in many engineering applications.Based on the research of the implementation process and characteristics of gravity search algorithm(GSA),several improvement strategies are put forward to solve the different array synthesis problems.The improved algorithms are used to optimize the benchmark function tests as well as the different target array synthesis problems.Comparing with other heuristic algorithms,effectiveness of the advised algorithms in the paper are verified through the corresponding experiments,design results and data analyses.The main contributions of this dissertation are summarized as:1.A hybrid gravity search algorithm(HGSA)is proposed.The "elite particles" and inertial mass regulation coefficient q are embedded into GSA.Firstly,in every iteration process of the algorithm,some excellent individuals with great potential may be eliminated.If such particles can be retained to next generation,the diversity of the population can be protected and the optimization accuracy can be improved,in this paper,the "elite particles" which belongs to potential particles are judged and retained to the next generation,the particles with poor fitness values are replaced by the "elite particles".Meanwhile,the inertia mass regulation coefficient q is designed in the GSA and used to calculate the inertia mass of the population,which widens the gap in space,speeds up the speed of particles approaching the optimal position in space and improves the convergence speed of the algorithm.The HGSA is used to synthesize low sidelobe linear array with different dimensions and phase-only pattern reconfigurable arrays.The optimization results verify that the HGSA proposed is superior to the standard GSA in both convergence speed and optimization accuracy under the same conditions,as well as the similar high-performance algorithm--moth flame-fighting optimization(MFO)algorithm.2.The interpolation gravity search algorithm(IGSA)is proposed.The IGSA can optimize the complex antenna arrays better such as the ultralow sidelobe and the low sidelobe with notches.Besides the adjustment coefficient q,the simplified quadratic interpolation(SQI)is embedded.The q is employed to change inertia weight distribution,the simplified quadratic interpolation(SQI)is served as a local search algorithm,it can enhance the local search ability of the algorithm,so the global and local search ability of the algorithm can be further balanced.The IGSA and multiple optimization algorithms are used to optimize the 13 benchmark test functions which contains not only some unimodal functions but also multi-peak multimodal functions,the proposed algorithm is also employed to synthesize different pattern arrays,based on the directivity diagrams,the convergence speed curve and the results of design objective including the main lobe width and sidelobe level,the better performance of IGSA is verified.3.In order to ameliorate the influence of random initial values on the convergence speed,the quasi opposition-based learning(QOBL)mechanism with adjustable probability is designed and applied to GSA,the quasi opposition-based learning gravity search algorithm(QOGSA)is proposed: according to different requirements of opposition operation in different stages of the algorithm,the times of QOBL are adjusted adaptively,so as to optimize the balance of global exploration ability and local exploitation ability of the population.Meanwhile,in order to offset the decline of population diversity caused by frequent opposition learning,elite particles are retained in the iteration process.QOGSA is applied to the optimization process of multiple test functions,the average best values are better than that of the improved algorithm exited which has a fixed opposition probability.Compared with GSA and the exited improved algorithm,QOGSA has best optimization performance to design different types of beams,so the exploration capability and exploitation capability can be better balanced.4.An adaptive gravity search algorithm(AGSA)is proposed to solve the impact on the exploitation capacity of GSA algorithm due to the almost same particle mass and force neutralization in the later iteration.An adaptive attenuation factor is designed which varies with the number of iterations;In addition,in order to improve the exploration ability of the GGSA algorithm,the influence of elite particles is increased in the process of speed change,so as to improve the global memory ability of the particles in the algorithm.The AGSA is applied to the synthesis of uniform circular array,sparse circular array,rectangular lattice planer array and sparse rectangular planer array.Compared with other improved algorithm,according to the optimization results of directional graph and convergence curve,it can be seen that the AGSA algorithm proposed has better effect in the synthesis of relevant planer arrays.