In the brain-machine interface (BMI) system, improving the motor recovery function of the joint and enabling the joint to adequately track the target trajectory are our main concerns. In this paper, to further improve the motor recovery effectiveness, two tasks are completed. First, a classical single-joint information transmission (SJIT) model is analyzed and improved by introducing several neurons. Second, an auxiliary controller that provides control inputs to the cerebral cortex is designed based on the model predictive control (MPC) strategy and is used to formulate a closed-loop motor rehabilitation BMI system based on the improved model. The simulation results show that the improved model can more accurately track the target movement trajectory than the SJIT model, and the designed BMI system can adequately recover the motor function of the joint. The sum of squared error (SSE) between the desired joint position trajectory and the output joint position trajectory is only \(1.2842 \times 10^{-5}\). In addition, under noises and disturbances, the BMI system can well recover the motor function of the joint. The model improvement and the designed BMI system are both effective.

在脑机接口（BMI）系统中，改善关节的运动恢复功能并使关节能够充分跟踪目标轨迹是我们的主要关注点。为了进一步提高电机的恢复效率，本文完成了两项任务。首先，通过引入几个神经元来分析和改进经典的单关节信息传输（SJIT）模型。其次，基于模型预测控制（MPC）策略设计了向大脑皮层提供控制输入的辅助控制器，并基于改进的模型来构建闭环运动康复BMI系统。仿真结果表明，与SJIT模型相比，改进后的模型可以更准确地跟踪目标运动轨迹，设计的BMI系统可以充分恢复关节的运动功能。所需关节位置轨迹和输出关节位置轨迹之间的平方误差总和（SSE）仅\（1.2842 \ times 10 ^ {-5} \）。此外，在噪声和干扰下，BMI系统可以很好地恢复关节的运动功能。模型的改进和设计的BMI系统都是有效的。