Neural network learning device, learning method, and MPPT converter control device using the same

摘要

PROBLEM TO BE SOLVED: To provide a learning device for neural network that can be mounted on a built-in system.SOLUTION: It is determined in S103 by additional learning of input whether the necessary number of neuron units is equal to or less than, or exceeds an upper-limit value of the number of units which can be mounted on a GRNN. When it is determined that the necessary number is equal to or less than the upper-limit value (Yes: S103), a new neuron unit is added to the GRNN in S105. When it is determined that the necessary number exceeds the upper-limit value (No: S103), on the other hand, a redundant unit is selected out of a plurality of neuron units constituting the GRNN in S107, and after the redundant unit is removed in S121, a neuron unit is added to the GRNN to replace the redundant unit.