The Application of Support Vector Machine in the Hysteresis Modeling of Silicon Pressure Sensor

Abstract

The diffused silicon pressure sensor is the mechanical-electrical-hydraulic system, so the pattern of the hysteresis is extremely complex. Because the Preisach model based on phenomenology is not limited to the particular physical nature and its hysteresis operator almostly describes any hysteresis, author studies the sensor hysteresis modeling with the Preisach model. The Preisach model can be obtained by regression analysis from the experimental data, which are acquired in the pressure calibration experiment of the diffused silicon pressure sensor. Because the sample from experimental data has the characteristic of the nonlinearity and the number of the samples is small, the author proposes to do regression analysis with support vector machine (SVM). Compared with the two-dimension regression analysis and BP neural network, SVM can achieve the more precise Preisach model rapidly.