The Prediction of Remaining Useful Life of Aluminum Reduction Cells Based on Improved Hidden Semi-Markov Model

Impact Factor:2.533

DOI number:10.1007/s11837-023-05703-y

Journal:JOM

Abstract:An improved prediction algorithm of the hidden semi-markov model (HSMM) is proposed to predict the remaining useful life (RUL) of aluminum reduction cells (ARC). First, the degradation process of the ARC is analyzed, and the parameters that could characterize its degradation process are determined. Second, to facilitate the prediction of HSMM, exponential distribution, in which form the dwell time in the traditional HSMM is distributed, is replaced with the Erlang distribution based on the queuing theory. Third, an improved forward recursion algorithm that introduced dwell time has not only simplified the calculation of useful life predictions but also further facilitated HSMM in the prediction of the RUL. Finally, verification is carried out using the actual data of an aluminum electrolytic industry, and the results showed that the improved HSMM performed better in the prediction and is more accurate than the existing prediction method.

Indexed by:Journal paper

Discipline:Engineering

Document Type:J

Volume:75

Issue:2

Page Number:2054-2063

Translation or Not:no

Date of Publication:2023-01-01

Included Journals:SCI