Penn State University's Applied Research Laboratory (ARL) engineers have developed a new multi-functional battery monitor that takes the physical condition of the battery and its age or past history into account as well as the percent of charge, resulting in highly accurate estimates of working time, physical condition and causes of failure based on a broad database.
The new monitor uses three automated reasoning methods, neural nets, fuzzy logic and auto-regressive moving average, and a decision block to combine the three estimates into a final prediction. The approach broadens the applicability of the monitor, since different reasoning methods perform better with different batteries, and enhances the reliability of the results.
James Kozlowski, ARL research assistant and doctoral candidate in electrical engineering who led the development team, said, "The new monitor is not limited to computer batteries but can be used in other battery-powered cordless or portable devices; car, boat and airplane batteries; and to monitor batteries in backup systems used in security, banking, communications and medical applications. It can monitor new or used, rechargeable or non-rechargeable, large or small batteries. It takes only 10 seconds to provide a reading and requires so little power, less than half a watt, it can be run off the power left in the battery you're testing."
The current prototype was built with off-the-shelf components and costs about $150.
