A multi-objective optimization algorithm based on self-organizing maps applied to wireless power transfer systems

In this work, a new multi-objective population-based optimization algorithm is presented and tested. In this contribution, the concepts of fast non-dominating sorting and density estimation using the crowding distance are used to create a multi-objective optimization algorithm based on previous work, which is a single objective evolutionary optimization algorithm based on self-organizing maps (SOMs). The SOMs paradigm introduces a strong collaboration between neighbors solutions that improves exploitation. Furthermore, the representative power of the SOMs enhances the exploration and diversification. A state of the art benchmark approach is used to evaluate the performance of the proposed algorithm, obtaining positive results. The test problem uses an analytical model of an inductively coupled wireless power transfer system (WPT). The objective is to optimize the WPT model characteristics in order to allow simultaneous data and power transfer between the coils. The WPT design approach uses more degrees of freedom than existing techniques leading to a number of solutions where both the power signals and the data signal can coexist on the same physical channel achieving good figures of merit.