Soft sensor design for a sulfur recovery unit using a clustering based approach

In the paper a Soft Sensor design strategy for an industrial process, via neural NMA model, is described. A general design strategy, based on the automatic selection of regressors of a NAM model is proposed. It is based on the minimization of the cost function of a Gath Geva clustering algorithm. The obtained soft sensor will be implemented in a refinery in order to replace the measurement device during maintenance to guarantee continuity in the monitoring and control of the plant. ©2008 IEEE.