A LS-SVM Based Nonlinearity Compensator for Thermocouple Vacuum Gauge

摘要

The input and output characteristic of thermocouple vacuum gauge exhibits high nonlinearity, which comes from the gauges' inherent physical peculiarities and causes big error in measurement results. There is a need to make compensation of the nonlinearity for accurate measurement. In this article, a least squares support vector machine (LS-SVM) model is developed to function as a compensator with which to accomplish this task. Particle swarm optimization (PSO) algorithm is introduced to search optimal parameters combination of the LS-SVM model. Based on the real sensor measurements obtained from a solar water heater plant, experiments were performed to investigate the compensation accuracy with developed LS-SVM compensator model and higher compensation accuracy results were obtained.