Parameter auto-selection for hemispherical resonator gyroscope's long-term prediction model based on cooperative game theory
作者：
Highlights：
• We propose a novel approach CoG-ARGM to auto-select optimal parameters based on the cooperative game theory.
• We map parameter auto-selection of CoG-ARGM to a transferable utility (TU) cooperative game and proof the game is convex.
• CoG-ARGM selects optimal parameters without searching all the parameters, reducing time consumption of searching.
• We put forward a reliability evaluation based on failure mode of FMEA to measure model's real-time prediction reliability.
• Experiments show CoG-ARGM yields better MAPE, Ln(Q), and time consumption compared with the state-of-the-art methods.
摘要
•We propose a novel approach CoG-ARGM to auto-select optimal parameters based on the cooperative game theory.•We map parameter auto-selection of CoG-ARGM to a transferable utility (TU) cooperative game and proof the game is convex.•CoG-ARGM selects optimal parameters without searching all the parameters, reducing time consumption of searching.•We put forward a reliability evaluation based on failure mode of FMEA to measure model's real-time prediction reliability.•Experiments show CoG-ARGM yields better MAPE, Ln(Q), and time consumption compared with the state-of-the-art methods.
论文关键词：Parameter optimization,Cooperative game theory,Hemispherical resonator gyroscope,Long-term prediction,Prediction reliability
论文评审过程：Received 28 September 2016, Revised 17 July 2017, Accepted 19 July 2017, Available online 22 July 2017, Version of Record 13 September 2017.
论文官网地址：https://doi.org/10.1016/j.knosys.2017.07.022
