Due to recurrent neural network’s (RNN) powerful ability to process sequential data, RNN attracts much attention in various fields. However, the long-term dependency has always been a major challenge for RNN. In this paper, a homotopy gated recurrent unit (H-GRU) model is proposed for predicting hyperchaos, which tries to improve the long-term dependence of RNN. We develop the model in three tasks: Mackey–Glass ($\tau$ $=$ 30), hyperchaotic Rössler, and 4D hyperchaos of Chen et al. (2018). Compared with the four pre-existing prediction models, the prediction accuracy of H-GRU is the highest among baseline models, which demonstrates the merit of the proposed model. Furthermore, the proposed model has significant improvement in replicating hyperchaotic attractors. Especially in 4D hyperchaos tasks, not only is the prediction accuracy of H-GRU three orders of magnitude higher than the baseline model, but also H-GRU can replicate hyperchaotic attractors more accurately.

由于递归神经网络（RNN）处理序列数据的强大能力，RNN在各个领域都备受关注。然而，长期依赖一直是 RNN 面临的主要挑战。在本文中，提出了一种用于预测超混沌的同伦门控循环单元（H-GRU）模型，该模型试图改善RNN的长期依赖性。我们在三个任务中开发模型：Mackey-Glass ($\tau$ $=$30），超混沌罗斯勒和 Chen 等人的 4D 超混沌。（2018 年）。与已有的四种预测模型相比，H-GRU 的预测精度是基线模型中最高的，这证明了该模型的优点。此外，所提出的模型在复制超混沌吸引子方面有显着改进。尤其是在 4D 超混沌任务中，不仅 H-GRU 的预测精度比基线模型高三个数量级，而且 H-GRU 可以更准确地复制超混沌吸引子。