Inverse Boundary Conditioning of Microsystems Through Fuzzy System Identification

The influence of geometrical variations, fabrication methods, fabrication process parameters and operational environment on the elastic behavior of microsystems is very important to be studied through the application of the concept of boundary conditioning. Considering the nature of complications at the microlevel, it may not be possible to predict the dynamic behavior under the influence of many boundary conditioning parameters and it may be necessary to develop the system from a priori knowledge. Hence, fuzzy clustering technique is applied to microsystems in this section. For simplicity, SISO (single input and single output) modeling is considered just to demonstrate the application of fuzzy system identification to model the inverse boundary conditioning in microsystems. The boundary conditioning of the MEMS plate type structure is considered for the present modeling. The input antecedent is the rotational stiffness at the stepped-up ends and the consequent is the first natural frequency. For the case of inverse modeling, the reasoning of generalized modus tollens was used along with Mamdani reasoning (Min-Max). Using GMT reasoning and the knowledge base, it was possible to predict the rotational stiffness required for a given natural frequency. The paper presents both model and predicted values. The comparison shows a good agreement. The paper also presents an example to demonstrate the principle. The example presented validates the use of fuzzy clustering with Mamdani inference mechanism for GMT reasoning