Design centering of analog circuits using GA and the regionalization method

In this paper, a new technique for design centering and feasible region estimation of analog filter circuits is introduced. The technique integrates genetic algorithms and the method of regionalization in a two-staged process of feasible region and design centering determination. Approximate boundary points of the feasible region are identified by a relaxed genetic algorithm (GA) search, and thus an estimation of the marginal boundaries of the feasible region is obtained. Using the method of regionalization, the marginal feasible range of each of the design parameters is divided into segments. The weight of each of the segments would depend on the its distance from the center of the design and the marginal probability density function is thus obtained. The joint probability density of the entire design can then be obtained by combining the marginal distribution functions. The technique is applied to the design centering problem of an analog filter circuit.