提出了一种基于傅里叶级数的非线性学习控制算法．当系统的任务可重复或系统的理想输出为周期函数时，在傅里叶空间内可以将时域内的轨迹跟踪问题简化为多个独立的调节器问题，因而大大简化学习控制器的设计．在滚珠丝杠工作台上的实验表明，这种学习控制器具有很快的收敛速度和令人满意的性能 A nonlinear learning control algorithm based on Fourier series is proposed. When the task of the system is repeatable or the ideal output of the system is a periodic function, the trajectory tracking problem in the time domain can be reduced to multiple independent regulator problems in the Fourier space, thus greatly simplifying the learning controller design. Experiments on the ball screw table show that this learning controller has fast convergence speed and satisfactory performance