Monotonic Degradation Electronics Life Analysis And Prediction Research

With the development of science and technology, electronic equipment has becomean important part of our daily life. Meanwhile, the accidents caused by the failure of theelectronic equipment are gradually increasing. In order to reduce these adverse effects,researchers are trying to find and replace the potential failure components in advance,which can avoid and manage system failures. Electronic equipment is composed ofvarious electronic components. Through predicting the remaining useful life of thesecomponents, consumers can replace the components which have reached the end-of-lifein time so as to avoid system failures. Lithium-ion batteries have been chosen as powersuppliers for many systems. Based on this situation, this paper chooses the lithium-ionbattery as an example to discuss the remaining useful life prediction methods. Themethods discussed in this paper are also suitable to other electronic components.This paper lists the application of extended Kalman filter, unscented Kalman filter,particle filter, unscented particle filter and unscented particle filter-Markov chainmonte carlo in the remaining useful life prediction of the lithium-ion battery. Firstly,describe the working principles and the structure of the lithium-ion battery, theoreticallyexplain the degradation reactions which cause the capacity decrease, and establish theequivalent circuit model and capacity degradation model of the lithium-ion battery.Secondly, apply the hidden Markov model to describe the dynamic system, establish thestate transition function and the measurement function, and combine with the capacitydegradation model so as to establish the lithium-ion battery capacity degradation systemmodel. When the measurements are known, if the state transition follows a first orderMarkov process, the state can be estimated by Bayesian theories. If the system model isnonlinear, the state can be estimated by the five algorithms described in this paper.Thirdly, introduce the conceptions and calculation steps of these five algorithms, takethe known capacity data into the system model to estimate the states, take the estimatedstates in the measurement function to predict the remaining useful life of the lithium-ionbattery. Finally, compare the remaining useful prediction results to analyze the methodsused in this paper. It is found that when the number of the known measurement data is small, particle filter–Markov chain monte carlo algorithm can effectively estimate theremaining useful life of the lithium-ion battery as well.Formerly, Kalman filter system methods and particle filter system methods werewidely used in target tracking, image processing, automatic control domain, and otherfields. The results in this paper show that these methods are also suitable to theremaining useful life prediction domain.