Prediction of Retained Capacity and EODV of Li-ion Batteries in LEO Spacecraft Batteries

Prediction of Retained Capacity and EODV of Li-ion Batteries in LEO Spacecraft Batteries S. Ramakrishnan, S. Venugopalan, A. Ebenezer Jeyakumar Abstract—In resent years ANN is widely reported for modeling in different areas of science including electro chemistry. This includes modeling of different technological batteries such as lead acid battery, Nickel cadmium batteries etc. Lithium ion batteries are advance battery technology which satisfy most of the space mission requirements. Low earth orbit (LEO)space craft batteries undergo large number of charge discharge cycles ( about 25000 cycles) compared to other ground level or space applications. This study is indented to develop ANN model for about 25000 cycles, cycled under various temperature, Depth Of Discharge (DOD) settings with constant charge voltage limit to predict the retained capacity and End of Discharge Voltage (EODV). To extract firm conclusion and distinguish the capability of ANN method, the predicted values are compared with experimental result by statistical method and Bland Altman plot. Keywords—Neural network, Lithium-ion Batteries, LEO Spacecrafts. ——————————  —————————— 1 INTRODUCTION Battery comprises complex set of interacting physi cal and chemical processes, the purpose of which is the conversion of chemical energy into electrical energy. The processes are often strongly influenced by the battery environmental conditions and usage profile. In fact, due to the number and complexity of the processes taking place, and the inability to accurately describe them, makes it difficult to develop an accurate battery model[1]. The Low Earth Orbit (LEO) spacecrafts are used for a wide variety of remote sensing applications such as
weather forecast, cartography, urban planning, environ- mental assessment, agriculture, forestry, ground water management etc. Batteries are required in LEO space- crafts for support during eclipse period and to meet peak power demand during sunlit. In LEO the batteries have to undergo 15 charge discharge cycles in a day. Typically the orbital period of a LEO spacecraft comprises 65 minutes of sunlit and 35 minutes of eclipse duration. Batteries are charged during the sunlit and discharged during eclipse to meet the spacecraft power demand. Batteries may be discharged even in sunlit to meet the peak power de- mand. In LEO the batteries are expected to last for 5 to 8 years. Consequently, LEO spacecrafts are highly demand- ing with respect to battery performance requiring 5500 charge discharge cycles in a year. Requirement of larger number of cycles forces to restrict the DOD, generally, to less than 30%. Lower the DOD, temperature and end of charge voltage (EOCV) limit, longer is the cycle life. In LEO orbits the batteries are used continuously without any rest or open circuit duration.

Lithium-ion batteries are high energy density rechargea- ble batteries that offer increased energy density, long cycle life, resistance to launch vibration, high reliability, wide temperature range of operation, radiation resis- tance, high round trip efficiency with no memory effect. These features of lithium-ion batteries lead to its selection in spacecraft programs. But this lithium-ion cell perfor- mance and the lifespan are largely dependent on parame- ters such as operating temperature, depth of discharge (DOD), and also charge discharge rates Since any failure of the batteries will lead to failure of spacecraft, they should be highly reliable and are to be tested extensively to stringent specification requirements before being put into use. The retained capacity of below 40% and EODV of less than 2.5V are considered to be failure of bat- tery[2][3].
Basically, mathematical models of physical systems are constructed to facilitate our understanding of mechan- isms that lead to specific responses and to enable re- sponse predictions .This creates the need for prediction tools that provide users with useful information such as remaining working time ,available energy at every de- sired time of operation, etc. In this regard, the Artificial Neural Networks (ANNs), one of the most powerful modeling techniques could be explored as a possible tool to predict the charge discharge characteristics of rechar- geable batteries. Because, ANNs play a vital role in ana- lyzing and predicting the behavior of systems that cannot be described by any analytical equations[4][5].
In recent years, multivariate methods and ANNs are used to predict the capacity behavior of lead-acid batteries
alone, wherein literature is replete with reports on the modeling and prediction of characteristics of lead-acid batteries . In this regard, ANNs have been used to predict capacity and power, gust effects on a grid-interactive wind energy

…(Full text truncated)…