Application of data-fusion compensating method based on FLANN constructed by SVM in CPS correction

Shiyuan Yang,Hua Dong,Dehui Wu
DOI: https://doi.org/10.3321/j.issn:0254-3087.2007.04.008
2007-01-01
Abstract:With the recognition of the construction method of generic functional link artificial neural network (FLANN), a novel construction method based on support vector machine (SVM) is discussed and applied to nonlinear correction and temperature compensation of capacitor pressure sensor (CPS). The topology construction of SVM was compared with that of generic FLANN, and their similarity was determined. Therefore, the optimum FLANN construction could be uniquely obtained by SVM through solving a quadratic programming. A generic FLANN was used to carry out a comparison experiment to solve the same problem. The simulation and experiment results show that this method has the characteristics of unique result, simple structure and global optimum, and it is strongly robust and highly accurate for a small sample in the case of insufficient experiment test data.
What problem does this paper attempt to address?