Parameters Adaption of Lebesgue Sampling-based Diagnosis and Prognosis for Li-ion Batteries

Wuzhao Yan, Wanchun Dou, Datong Liu, Yu Peng, and Bin Zhang
Submission Type: 
Full Paper
AttachmentSizeTimestamp
phmc_15_069.pdf507.44 KBAugust 25, 2015 - 8:26pm

Traditional fault diagnosis and prognosis (FDP) approaches are based on Riemann sampling (RS), in which the samples are taken and algorithms are executed in a periodic time interval. With the increase of system complexity, the real-time implementation of this RS-based FDP (RS-FDP) becomes a bottleneck, especially for distributed applications. To overcome this problem, a Lebesgue sampling (LS)-based FDP (LS-FDP) is proposed. LS-FDP takes samples on the fault dimension axis and provides a need-based solution, which implies a diagnostic philosophy of “execution only when necessary”. It is promising in computational cost reduction and uncertainty management. In the previous Lebesgue sampling-based prognostic philosophy [1], [2], the predefined fault state distance (Lebesgue length) is a constant, which implies that the Lebesgus events have the same priority in FDP. However, the severe change of system state usually means a high possibility of fault occurrence, the diagnostic and prognostic algorithms need to be executed more frequently under such circumstances to have a closer monitoring on the system state. Thus, the Lebesgue lengths are changed according to the slope of the system feature value, and result in a Lebesgue sampling-based prognostic method with weighted Lebesgue state in different system stages. The goal of this paper is to deliver an improved LS-FDP method with varying Lebesgue length which enable the FDP to be executed based on state variation speed. This method is an optimal solution for deploying FDP in limited computational resource system. The design and implementation of varying Lebesgue length LS-FDP based on particle filtering algorithms are illustrated with experimental results on Li-ion batteries to verify the improvements of the proposed approaches.

Publication Year: 
2015
Publication Volume: 
6
Publication Control Number: 
069
Page Count: 
9
Submission Topic Areas: 
Model-based methods for fault detection, diagnostics, and prognosis
Submitted by: 
  
 
 
 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed