Prognostics for automotive, marine and heavy industry
Chair: Prof. Bogdan I. EPUREANU (University of Michigan)
Research on the degradation model of the clamping device based on CAE simulation and Data-driven Ms. Li YU, Rui KANG, Bin ZHANG and Yansong WANG (MESNAC CO.,Ltd., Beihang University)
The paper takes the clamping device of curing machine as an example, the CAE simulation and Data-driven are combined to research the degradation model building. Based on analysis of the failure mode and failure mechanism, it is concluded that the reason of clamping device cracking was high cycle fatigue failure. The durability test was conducted by CAE simulation analysis, and the test data of the eigenvalue of the device have been collected by the NI data acquisition equipment. The research based on the collected data, using Data-driven to model the degradation caused by high cycle fatigue. And by the comparison of different regression models, we selected the appropriate one as Bayesian prior model, which will provide theoretical support for PHM management of similar machines. In applications, by updating this model with actual degradation data of the target device, can monitor the health state and predict the remaining service life of the target device. So that the actively predictive maintenance can be done and the failure can be avoided.
Research on Fault Detection Method for Tire Dynamic Balance Measuring System Based on Correlation Analysis Ms. Yueyue LIU, Xiaoyang LI, Deng LIU and Wenbin CHEN (MESNAC Co.,Ltd, Beihang University)
Application of correlation analysis method was studied in fault detection of dynamic balance measuring system. The test data reproducibility in dynamic balance measuring system is a key index of the system’s reliability, there are many influencing factors and all of them are with relevance,relevance,which make it difficult to troubleshoot the factors that affect the reproducibility of test data in practical project,this has been restricting the development of tire dynamic balance measuring systems. The method of correlation analysis was used to the fault detection for tire dynamic balance measuring system by the first time.The sensitive factors of test data repeatability in dynamic balance measuring system was obtained through the correlation analysis of the measured data in the measuring system, and the measuring system was optimized according to the analysis results, then a series of tests were arranged in the optimized system to verify the correctness of the correlation analysis,thus clarifying the important role of correlation analysis in fault detection.
Assembly Quality Diagnosis of Planetary Gear Sets Mr. Jiung HUH, Soonyoung HAN and Hae-Jin CHOI (Chung-Ang University)
Product quality is one of the most important factors to be considered in manufacturing industries. Autonomous production line rapidly increases its productivity; however, quality check becomes a difficult problem and a smart system for quality assurance is indispensable in the modern production line
In this study, we developed a transmission error based machine learning algorithm to check the quality of planetary gear assemblies and identify the defective parts in the planetary gear sets.
A phenomenological force model of Li-ion battery packs for enhanced performance and health management Dr. Ki Yong OH and Bogdan I. EPUREANU (University of Michigan)
A 1-D phenomenological force model of a Li-ion battery pack is proposed to enhance the control performance of Li-ion battery cells in pack conditions for efficient performance and health management. The force model accounts for multiple swelling sources under the operational environment of electric vehicles to predict swelling-induced forces in pack conditions, i.e. mechanically constrained. The proposed force model not only incorporates structural nonlinearities due to Li-ion intercalation swelling, but also separates the overall range of states of charge into three ranges to account for phase transitions. Moreover, an approach to study cell-to-cell variations in pack conditions is proposed with serial and parallel combinations of linear and nonlinear stiffness, which account for battery cells and other components in the battery pack. The model is shown not only to accurately estimate the reaction force caused by swelling as a function of the state of charge, battery temperature and environmental temperature, but also to account for cell-to-cell variations due to temperature variations, SOC differences, and local degradation in a wide range of operational conditions of electric vehicles. Considering that the force model of Li-ion battery packs can account for many possible situations in actual operation, the proposed approach and model offer potential utility for the enhancement of current battery management systems and power management strategies.