Application of the Runge-Kutta fourth order integration in Coulomb counting method for indoor drone battery state of charge estimation

This thesis explores the impact of numerical integration methods on State of Charge (SOC) estimation accuracy for drone battery monitoring, specifically comparing the Euler single-step method and the Runge-Kutta 4th order (RK4) method. Contrary to expectations, results on a static dataset with mostly linear current trends showed that Euler exhibited lower cumulative error than RK4. This outcome is attributed to the simplicity of the data, which favored the less complex integration approach. Despite RK4’s theoretical advantages in handling non-linearities, its

performance benefits did not materialize under the static test conditions. Furthermore, RK4 demonstrated approximately double the processing time of Euler, which may present challenges for real-time drone applications. However, RK4 remains a promising candidate for future implementation, particularly in real-time, curved, or dynamic current profiles where its higher-order accuracy is expected to reduce integration error over time. This work establishes a baseline for further exploration into live SOC estimation improvements and supports the case for

targeted integration method selection based on the data profile and system constraints.