An expert system for digital filter design

The objective of this research is to consider the feasibility of an expert system for designing digital filters. The prototype model serves as the foundation for an expert system capable of deriving a set of transfer function coefficients from a set of points that are sketched by the user. In this expert model, the user specifies a design by sketching (using a mouse) a curve which approximates the desired magnitude plot. The expert system then operates on the set of points which define the curve to determine the design parameters, compute the transfer function, and finally produce a magnitude plot. The current filter possibilities are designed as cascaded second-order section with real coefficients. The authors describe the characteristics of this expert system and focus on the techniques used for implementation and knowledge representation. A design example is included