【目的】 针对有源电力滤波器（active power filter， APF）谐波治理中的电流跟踪难题，提出一种基于海马体模糊神经网络（hippocampus-based fuzzy neural network， HBFNN）的快速积分终端滑模控制（fast integral terminal sliding mode control， FITSMC）策略。【方法】 利用FITSMC保障全局鲁棒性与误差有限时间收敛；构建HBFNN在线逼近系统未知动态以突破参数依赖。HBFNN融合海马体机制与模糊理论，通过特征选择剔除冗余，并利用双递归结构增强对时变信号的抗扰性能。【结果】 仿真与硬件实验验证表明，所提HBFNN-FITSMC方案可快速精确跟踪谐波电流。仿真中，电网侧电流总谐波失真（total harmonic distortion， THD）从40.30%降至1.25%；硬件实验中，THD从32.73%降至2.96%。与传统方法相比，该策略显著提升了动态响应与稳态精度，并有效抑制了系统抖振。【结论】 HBFNN凭借信息筛选与双递归机制，具有更优逼近与抗扰能力，降低了对精确模型的依赖。该方案实现了类脑智能与滑模控制的优势互补，具有显著的工程应用价值。

[Objective] To address the current tracking challenges in the harmonic suppression of active power filters （APF）， a fast integral terminal sliding mode control （FITSMC） strategy based on a hippocampus-based fuzzy neural network （HBFNN） is proposed. [Methods] The FITSMC is employed to guarantee global robustness and finite-time convergence of the tracking error. To circumvent the dependence on accurate system parameters， the HBFNN is constructed to approximate unknown system dynamics online. By integrating the hippocampus mechanism with fuzzy theory， the HBFNN eliminates redundancy through feature selection and enhances anti-interference performance against time-varying signals via a double recurrent structure. [Results] Simulation and hardware experiments verify that the proposed HBFNN-FITSMC scheme tracks harmonic currents rapidly and accurately. In simulations， the total harmonic distortion （THD） of the grid-side current decreases from 40.30% to 1.25%， while in hardware experiments， it decreases from 32.73% to 2.96%. Compared with traditional methods， the proposed strategy significantly improves dynamic response and steady-state accuracy， while effectively suppressing system chattering. [Conclusion] By virtue of its information screening and double recurrent mechanism， the HBFNN demonstrates superior approximation and anti-interference capabilities， reducing reliance on precise mathematical models. This scheme achieves the complementary advantages of brain-inspired intelligence and sliding mode control， offering significant value for engineering applications.