The Application Of Evolutionary Algorithm For Broadband Matching Network

To cope with highly nonlinear, nondifferentiable and multi-peak objective function, evolutionary algorithms which have the advantage of global solution searching can overcome the drawbacks of conventional algorithms such as slow speed, relatively poor precision and convergence to local solutions.For realizing the biggest transmission power in the useful pass band, an essential problem in the communication system is to design a lossless matching network between the source and the load. This process include two points, on one hand, the best topology of the network should be chosen; on the other hand, the element parameters should be optimized .GA can theoretically convergence to the global solution, but actually, it can not always avoid premature convergence for many conditions are not perfect. How to design an effective evolutionary algorithm for a certain problem is discussed in this paper.In the paper, a special binary-tree coding method is applied to coding the topology of the matching network and a simple clone selection algorithm to optimize the component parameters of the network is embedded in the main program. By comparing the design experiment with those known methods, the fact that the binary-tree-code genetic algorithm is an effective method for designing matching networks is testified.At the end of the dissertation, a summary is given and future wok is pointed out.