Feasibility Study of Thermal Prediction of Valve Regulated Lead Acid Battery

閥調式鉛酸電池溫度變化預測之可行性研究

Student thesis: Master's Thesis

View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date22 Jan 2021

Abstract

Valve regulated lead acid (VRLA) battery manifests plentiful applications and thus inherits a large market share in the industry. VRLA battery is widely adopted in data center, airplane,… etc as a result of its low cost and high adaptation in extreme conditions. However, VRLA battery is subjected to the risk of thermal runaway which potentially may result in battery damage and thus causing fire hazard and potentially serious accidents. It is obvious that the battery temperature has become a key parameter to be monitored. In my research, two effective VRLA battery temperature monitoring schemes have been developed. The first model is referred to as the adaptive lumped thermal model (ALTM). This model predicts the battery temperature by analyzing the chemical reactions inside the battery during in-charge or discharge. Eight representative parameters of the battery are identified for analysis. ALTM is the first of its kind, which does not require a prior knowledge of the battery structure. Above all, the developed model can be applied to VRLA batteries of different brands. The second scheme is developed, namely VRLA battery internal temperature prediction (VBITP) to predict the battery temperature according to the value of input current (IC) and ambient temperature (AT) from external sensors. In this scheme, nonlinear autoregressive exogenous (NARX) algorithm is customized to model the relationship between the input and the internal temperature (IT) (output). Compared with the conventional VRLA internal temperature monitoring method, the proposed schemes do not rely on the installation of thermocouples inside the battery, which greatly reduces installation and maintenance costs. Experiments results revealed that both two proposed schemes achieved high prediction accuracy with a root mean sequence error (RMSE) of better than 0.604℃ (ALTM) and 0.05℃ (VBITP). In the era of smart energy management, the proposed schemes would greatly reduce the cost caused by internal sensors and facilitate the efficient production of the VRLA batteries.