The aim of this thesis is to develop an accurate method for estimating state of charge (SOC) estimation of Li-Ion batteries and use it in the real-world setting in an intelligent mobile home. The state of charge can not be measured directly. While it depends on the past electrical current that flew in/out of the battery and the battery temperature, a reasonable estimation can be made based on the current values of the battery voltage, temperature and incoming/outcoming electrical current.
In the theoretical section, we present various battery technologies, and methods that can be used to predict the SOC information. The methods are divided into four categories: direct methods, integration methods, hybrid methods and adaptive systems. We describe the basic characteristics of each groups and of some methods.
In the practical section, we first acquired the historical measurements of the battery current, voltage and temperature under regular operating conditions. We use the data to compare suitable representatives of each groups of methods. We found out that the best performing method was a combination of the linear regression with voltage measurement and support vector machine, so this method was eventually implemented in the mobile home.
|
