Diesel engine generator is consist of diesel engine and generator. This is used for island and ship as power source and emergency generator of large buildings and plant. Diesel engine generator need to monitoring system for stable operation. Various sensors are used for monitoring the condition of the diesel engine generator. Due to limited battery life and space constraints, it is difficult to supply power to these sensors. One way to resolve this problem is to use an energy harvester. The casing of a diesel engine generator typically vibrates in the frequency range from 10 Hz to 20 Hz. Among vibration energy harvester, Electromagnetic energy harvester is suited to low frequency range. Electromagnetic energy harvesters consist of magnetic source and coil that are determined by such parameters as the size, magnetization pattern, and the number of turns.
This paper performs an optimization for designing an electromagnetic vibration energy harvester within a limited volume and conditions. For this optimization, we compared with morphological of electromagnetic energy harvest. The result show that magnet of planar type and halbach array, stator yoke are suitable for inducing high voltage. In order to fast calculate the flux or induced voltage with respect to the size and position of magnet and coil, we needed mathematical modeling. So we used analytical method with current sheet equivalent and image method. The mathematical modeling is very accurate(maximum 1% mismatched with FEA).
Based on mathematical modeling, we established cost function of optimization and constraint condition with frequency of 10Hz and 3.7V Li-ion battery. We performed optimization about such parameter as the coil space and coil thickness, number of magnet. The optimization result of electromagnetic energy harvester show that coil space is 9.5mm and coil thickness is 9mm, number of magnet is 5. To validate the optimization, we implemented electromagnetic energy harvester. The result of experiment, open circuit voltage is about 3.1 times mismatched with optimization. The reason is that induced current is not considered and damping of prototype is not correct. So power is about 25% mismatched with optimization.