Multivariable Support Vector Regression with Multi-sensor Network Data Fusion
Chan‐Yun Yang, Chen-Yu Lin, Sainzaya Galsanbadam, Hooman Aghaebrahimi Samani · 2018
Motivated by modeling a general behavioral function of a target system, a data-driven multivariable support vector regression (SVR) is developed. The multivariable SVR is sought to estimate a generalized relationship among multiple input variables which would be collected locally from a distributed multi-sensor network. Instead of an immediate estimation with a installed single sensor, the participation of distributed multi sensors gains the estimation of the system states more reliable and more generalized. The proposed SVR modeling method gives rise of the possibility to support multivariable input spectrum, and the flexibility to combine the input variables with respective decay weights. The paper herewith elaborates the development with experimental evidences.