Study On Invasive Weed Optimization And Its Application For Pattern Synthesis Of Antenna Arrays

The development of antenna arrays raises many difficult problems of pattern synthesis. These problems can be cast into multi-parameter, multi-restriction and non-linear optimization problems in terms of mathematics. This presents an enormous technical challenge. Intelligent optimization methods based on novel information processing mechanism and computational model have been considered and successfully applied to array pattern synthesis. Invasive weed optimization inspired from weed colonization is a new numerical stochastic optimization algorithm. It is shown that this optimizer not only in certain instances outperforms other existent optimizers, but also is capable of handling some array pattern synthesis problems. However,the research on invasive weed optimization and its application on array pattern synthesis is not enough. There are many problems to be studied, such as characteristic, potential, improvement and application of invasive weed optimization. Above-mentioned aspects are also main body of this paper. The major contributions of this paper can be summarized as follows:1. The characteristics of invasive weed optimization and particle swarm optimization are studied and digested. Particle swarm optimization is used to improve local search ability of invasive weed optimization, and a hybrid invasive weed optimization with particle swarm is designed.2. A hybrid self-adaptive invasive weed optimization is proposed to improve the optimization efficiency. Self-adaptive parameter control approach is adopted to avoid the difficulty of parameter selection. In addition, the mutation, crossover and selection operator in differential evolution algorithm are incorporated in the proposed method to enhance its explorative power.3. The crossover operation in genetic algorithm is used to enhance the search ability of binary invasive weed optimization. A hybrid binary invasive weed optimization is designed.4. Typical instances of test function are presented to illustrate the performance of the proposed method.5. Then, above-mentioned algorithms are respectively used to produce null and depress side-lobe for unequally spaced arrays and thinned arrays. Effectiveness of new methods is validated by means of experiments.