With the appreciating of problems related to the environmental impact evaluation of human activities, wireless sensor networks (WSNs) are emerging as suitable monitoring systems for large geographical areas, fixing new challenges in many scientific areas such as computer science, wireless communication and energy harvesting. In recent years tubular permanent magnet linear generators (TPM-LiG) have been proposed in several marine applications as a well-suited technology for power generation such as power buoys. In this work an automated soft-computing approach is presented in order to provide an optimized electromagnetic model of TPM-LiG. A prototype of tubular generator was set up and measures provided, to a large extent, a first validation of the predictive model used in the proposed automated approach.