In order to solve the problem of parameter setting of the extended state observer used in nonlinear dynamic system,a parameter setting model based on BP neural network is established.The online gradient descent method is used to train the network so as to ensure the learning ability of the dynamic system.The incremental Delta-Bar-Delta algorithm is introduced,and the information of input data and learning experience are used to realize the adaptive adjustment of the learning rate and improve the adaptability of the online gradient descent method.It can be seen from the numerical simulation that the parameter setting model possesses the advantages of good dynamic performance and high accuracy compared with the traditional parameter setting model,and it can improve the dynamic setting accuracy of the parameters of the extended state observer of the nonlinear system,so that the controlling performance of the ADRC system is improved to a certain extent.