Multiple correlation analysis for finite-time delay estimation in Soft Sensors design

Measurement finite-time delay is a relevant problem in the process industry. A procedure for its estimation in the case of real industrial data is proposed and verified by using a set of synthetic dynamic case studies. Results referring to an industrial case study are also reported. The procedure faces the effects of input mutual correlation by using the squared multiple correlation index. Reported results show the suitability of the proposed procedure in solving the investigated class of problems. This allows the design of data-driven Soft Sensor for the real-time estimation of product quality variables in industrial applications.