针对实际非线性、不确定系统,文章将提出基于BP算法神经网络PID算法,阐述了三层BP网络设计方法。最后用于一个慢变性被控对象进行仿真,结果表明,此控制方法能够有效克服经典PID控制器在被控对象具有非线性、时变不确定性和难以建立精确的数学模型时出现的参数整定不良等问题。 In view of the actual nonlinear and uncertain system, the paper proposes a neural network PID algorithm based on BP algorithm, and expounds the design method of three-layer BP network. Finally, a slowly degenerate controlled object is simulated. The results show that this control method can effectively overcome the parameters tuning of classic PID controller when the controlled object has nonlinearity, time-varying uncertainty and difficult to establish accurate mathematical model Bad issues.