Automatic Filter Design Algorithm based on Machine Learning and Synthesis

The traditional Film Bulk Acoustic Resonator (FBAR) filter design process is usually assisted by optimization algorithms. The design process requires multiple iterations of parameters in a large range to reach convergence, resulting in high time cost and low design efficiency.A algorithm combining machine learning and circuit synthesis is proposed in order to improve the design efficiency of FBAR filters. Through this algorithm, the structure of each FBAR can be obtained by only inputting the sampled value of S parameter, instead of iterating the structure of FBAR in the conventional algorithm.The filter constructed by this algorithm has better performance. At the same time,the design time is much shorter than the traditional algorithm, thus the design efficiency can be significantly improved.