Two-stage fuzzy inference system for symbolic simplification of analog circuits

This paper presents a knowledge-based fuzzy approach to symbolic circuit simplification in an effort to imitate human reasoning and knowledge of circuit designer experts. The fuzzy approach differs from the conventional simplification techniques in that it can efficiently combine different input variables to obtain optimal simplified expressions. Additionally, this method was chosen due to its adjustability and interpretability, as well as its ability to manage very complex symbolic expressions. The proposed algorithm uses fuzzy logic to simplify the symbolic circuit transfer functions in two stages. In the first stage, a fuzzy system is applied to directly eliminate nonessential circuit components, resulting simplified circuit topology which also yields simpler transfer function. In the second stage, another fuzzy system is used to further simplify the symbolic transfer function from the already simplified circuit, such that deeper insight into the circuit behavior can be obtained. Symbolic and numerical results show that the fuzzy approach outperforms the conventional techniques in terms of accuracy, expression complexity, and CPU running time.