23-06-2023 дата публикации
Номер: CN116306838A
Принадлежит:
The invention belongs to the technical field of neural network optimization, and particularly relates to a KELM neural network optimization method based on an improved squirrel algorithm, which is characterized in that squirrel populations are classified through a fitness threshold value in an iteration process, and different dynamic sliding coefficients are respectively adopted, so that the squirrel populations are optimized; in this way, the early-stage global search capability and the later-stage local exploration capability of the algorithm are enhanced. And the non-high-quality squirrels are subjected to large-scale global optimization, and the high-quality squirrels are subjected to small-scale fine exploration, so that the optimization performance of the algorithm is improved. An information sharing strategy is adopted to update the positions of squirrels on a pecan tree, high-quality information of a population is effectively utilized, the convergence speed of an algorithm is increased ...
Подробнее