Battery Charge And Source Estimation As A Product Component
This paper's goal is to present a low cost, non-conventional solution for battery state of charge estimation and external electrical input presence/absence for a commercial mobile, handheld device whose battery state of charge control is critical. This solution is based on treating and filtering a time series in real-time software, using the battery pack characteristic discharge curve and time series statistical features. The time series is composed of data that is sampled embedded in hardware, communicating directly with the machine's BIOS. The system processes this data and outputs a value that indirectly relates to state of charge, needing further processing to insure accuracy. The data stream is treated in a process that directly relates the output time series with state of charge through a transfer function, effectively treating intermediary conversions as black boxes to simplify analysis and implementation. This process can also detect if an external source is connected/disconnected by exploiting pre-detected features in the time series. This approach advantages are its low cost and simplicity, reducing hardware complexity and expenses; small dimensional footprint; mostly software-based; and centralization into the main hardware as low computational cost daemons, simplifying data consumption.