Research On Parameters Extracion Of Filters With Limited Quality Factor

The traditional optimization and diagnosis wastes time and energy in the design progress of microwave filter. But we can greatly shorten the design period by researching the parameters of the filter to guide the optimization. So the parameters extraction technology has received wide attention.The coupling matrix is the math expression of the filter. The coupling parameters of the matrix has a closed relationship with the physical dimensions of the fitler. So we can extract the coupling matrix and utilize the mapping relationship of the coupling matrix and the physical dimensions to direct the direction and range of the optimization of filter. At present, the parameters extraction technology of the lowloss filter has formed a unitary system.However, to the limited quality factor filter such as mircostrip and SIW filters, the parameters extraction has a bad result, related to the low-loss filer, such as cavity filter and SIW filter. The reason is that the coupling matrix is a lossless model, but he filter in reality is a lossy model. The two models always have differences. At present, there are several types method to extract the lossy filter’s coupling matrix, they adopt different function to approximate the filter response, and the extraction results are different and not good. So it is necessary to find a valid method to extract the parameters of the limited quality factor filter and put forward a standard to assess the effectiveness of different mothods. In this paper, we proposed a method based on the equivalent circuit model and curve-fitting technology to extract the coupling matrix of the lossy filter. The contend of thesis are as follows:(1)Firstly, introduced the synthesis theory of the coupling matrix and the parameters extraction technology.(2) Then, based on the theory of coupling matrix of lossy filter, thesis improved the equivalent circuit model and use the model to extract the coupling matrix of the filter sample.(3) Then used the neural network to build the mapping between the coupling matrix and physical dimensions. And then thesis design a micro strip CT filter to test the method we improved.