Battery Charge Controller Using Artificial Neural Network-PI (ANN-PI) Algorithm for Various Voltages Rate

Seen from track record the battery development is classified as very rapid. Batteries are needed by the majority in the field of renewable energy, electric vehicles, and other electrical devices. However, to reduce battery waste which will cause new problems, there are now many batteries circulating rechargeable instead of disposable batteries. Each battery rechargeable has various specifications. These specifications include voltage charging, battery capacity, and also current charging. The battery has a variety of specifications, each battery has an adapter charger different too. Therefore, this research created a system of battery charging controller which can charge the battery with different specifications without many adaptors. On the system is used SEPIC (Single Ended Primary Inductance Converter) Converter with settings duty cycle to generate voltage output which varies. There are four variations of the battery with different specifications. Charging is not done simultaneously, only one battery only. Output current and output voltage will be used as input algorithm Artificial Neural Network (ANN) to perform battery identification and determination set point. Then, PI control stabilize the output in the form of set points. And the method of charging which is used on the system is constant voltage. The system will provide an output voltage of 28.8volt for charging a 24volt battery, 14.4volt for charging a 12volt battery, 7.2volt for charging a 6volt battery, and 8.88volt for charging a 7.4volt battery. Obtained, the average output voltage error is 1.045% for the PI control and 0.13% for the ANN-PI control method for charging on four batteries with varying voltages.