A graph-guided network with adaptive evaluation and improvement for disturbed sensors in fault-tolerant soft sensor modeling

Liyuan Kong, Chunjie Yang, Siwei Lou, Yaoyao Bao, Xiaoke Huang, Li Chai

Abstract: Highlights•A fault-tolerant soft sensing model relaxes the requirement for data quality.•This paper comprehensively considers instantaneous noise and continuous drift.•Point-series deviations are identified and processed in a differential manner.•The quality of disturbed data is enhanced against noise and drift.•This method’s advantages are well demonstrated in a real cement production process.