Abstract:Summary
The state of charge and state of health estimations are two of the most crucial functions of a battery management system, which are the quantified evaluation of driving mileage and remaining useful life of electric vehicles. This paper investigates a novel data‐driven–enabled battery states estimation method by combining recurrent neural network modeling and particle‐filtering–based errors redress. First, a recurrent neural network with long‐short time memory is employed to learn the long‐term nonlinea… Show more
“…Advances in lithium‐ion battery technology make it ubiquitous in battery powered devices, ranging from electric terrestrial and aerial vehicles to consumer electronics. Therefore, battery health assessment becomes a crucial problem in providing safety and reliability of the system 1,2 …”
Summary
Online state of health (SOH) prediction of lithium‐ion batteries remains a very important problem in assessing the safety and reliability of battery‐powered systems. Deep learning techniques based on recurrent neural networks with memory, such as the long short‐term memory (LSTM) and gated recurrent unit (GRU), have very promising advantages, when compared to other SOH estimation algorithms. This work addresses the battery SOH prediction based on GRU. A complete BMS is presented along with the internal structure and configuration parameters. The neural network was highly optimized by adaptive moment estimation (Adam) algorithm. Experimental data show very good estimation results for different temperature values, not only at room value. Comparisons performed against other relevant estimation methods highlight the performance of the recursive neural network algorithms such as GRU and LSTM, with the exception of the battery regeneration points. Compared to LSTM, the GRU algorithm gives slightly higher estimation errors, but within similar prediction error range, while needing significantly fewer parameters (about 25% fewer), thus making it a very suitable candidate for embedded implementations.
“…Advances in lithium‐ion battery technology make it ubiquitous in battery powered devices, ranging from electric terrestrial and aerial vehicles to consumer electronics. Therefore, battery health assessment becomes a crucial problem in providing safety and reliability of the system 1,2 …”
Summary
Online state of health (SOH) prediction of lithium‐ion batteries remains a very important problem in assessing the safety and reliability of battery‐powered systems. Deep learning techniques based on recurrent neural networks with memory, such as the long short‐term memory (LSTM) and gated recurrent unit (GRU), have very promising advantages, when compared to other SOH estimation algorithms. This work addresses the battery SOH prediction based on GRU. A complete BMS is presented along with the internal structure and configuration parameters. The neural network was highly optimized by adaptive moment estimation (Adam) algorithm. Experimental data show very good estimation results for different temperature values, not only at room value. Comparisons performed against other relevant estimation methods highlight the performance of the recursive neural network algorithms such as GRU and LSTM, with the exception of the battery regeneration points. Compared to LSTM, the GRU algorithm gives slightly higher estimation errors, but within similar prediction error range, while needing significantly fewer parameters (about 25% fewer), thus making it a very suitable candidate for embedded implementations.
“…Several DBM-AT models have been proposed for both calendar aging and cycling aging modes and different cell chemistries, mostly using the capacity fade as a health indicator [11][12][13][14][15][16][17] ; we will turn our attention first to calendar aging. Schmalstieg et al 18 performed calendar aging tests on 2.15-Ah lithium nickel manganese cobalt (NMC) oxide cells, and they fitted the data of capacity loss as a function of temperature, SOC, and time using a set of quadratic equations.…”
Summary
This work summarizes the findings resulting from applying an aging modeling approach to four different capacity loss experimental datasets of lithium‐ion batteries (LIBs). This approach assumes that the degradation trajectory of the capacity is a function of three variables: time, kinetic constant, and time‐dependent factor. The analysis shows that the time‐dependent factor α is cell‐chemistry dependent and cannot be averaged for calendar and cycling modes and combined modes. This factor was also found to be a function of the stress factors. A quadratic model was used to obtain the kinetic constants per test, and statistical metrics were provided to evaluate the quality of the fitting, which was significantly affected when using averaged values of α and refitted kinetic constants. A set of test matrices is proposed for calendar, cycling, and mixed aging modes to overcome the challenges of data‐based models developed from accelerated test approaches for modeling aging in LIBs. This work also proposes a methodology to develop these data‐based aging models.
“…Therefore, this method is difficult to apply in EVs. Besides, there are some machine learning methods represented by Support Vector Machines, Fuzzy Logic, Gaussian Process Regression, and Neural Networks . This method does not need a battery model that characterizes the complex internal principles of battery operation .…”
Summary
A state of health (SOH) estimation method that can be achieved online and only requires battery management system (BMS) detection data is proposed in this article. In the State of Health mathematical model proposed in this article, the using time of power battery is treated as an independent variable and SOH is treated as a hidden variable. And the mathematical model just used online process data from BMS. So it would make the SOH estimation method more suitable for actual engineering. Then, the article proposes an interleaved time model parameter update framework to reduce the computational complexity of the algorithm in a single sampling period. In this framework, we propose a fast model parameter identification algorithm that uses nonlinear least squares to initialize a genetic algorithm searched range. Finally, the whole method is verified by using the NASA database. The results prove that the proposed online SOH estimation method has higher SOH estimation accuracy and is more suitable for engineering applications in the field of electric vehicles than the existing SOH estimation methods.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.