We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
THE VIBRATION SOFT SENSING MODEL OF CERAMIC SPINDLES BASED ON LEAST SQUARES SUPPORT VECTOR MACHINES.
- Authors
Aoxiang LIU; Zinan WANG; OANCEA, Gheorghe; Xiaotian BAI; Zhipeng WU; Qi WANG
- Abstract
The prediction and analysis of ceramic motorized spindle vibration characteristics are of great significance due to their high accuracy within a short time frame, especially in a strong corrosion, ultra-high temperature extreme environment. This paper establishes a soft sensing modeling method based on least squares support vector machines (LSSVM) for the vibration of a ceramic spindle. The shell temperature of the spindle optimal point was analyzed, combined with the selection of different rotation speeds. Then, the soft sensing model of ceramic spindle vibration was designed by LSSVM. The prediction performance of the vibration model was compared with the Partial Least Square (PLS) and Back Propagation (BP) neural network models, respectively. The comparison results reflect that the spindle prediction model developed through LSSVM has a geart prediction performance through Maximum Error (ME), Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) criterion than PLS and BP neural networks, with the highest prediction accuracy. The model obtained higher ceramic spindle vibration prediction accuracy, and the feasibility and effectiveness of the LSSVM method were validated. The LSSVM method for ceramic spindle prediction can provide theoretical guidelines for petrochemical, aerospace, and mechanical processing factory researchers and users.
- Subjects
SLEEP spindles; SUPPORT vector machines; LEAST squares; ARTIFICIAL neural networks; STANDARD deviations; BACK propagation
- Publication
Academic Journal of Manufacturing Engineering, 2021, Vol 19, Issue 2, p90
- ISSN
1583-7904
- Publication type
Article