ＢＰ网络是目前应用很广的人工神经网络。改善ＢＰ网络算法尤其是提高ＢＰ网络的学习速度一直是人们研究的重点课题。本文在对ＢＰ网络现行算法研究的基础上，提出根据系统误差自适应调节学习率的快速ＢＰ算法（ＦＢＰ），并以棉花棉铃虫预测为例，对两种算法的收敛速度进行了比较。 BP network is a widely used artificial neural network. Improving BP network algorithm especially to improve the learning speed of BP network has always been a key issue for people to study. Based on the research of the current algorithms of BP network, this paper proposes a fast BP algorithm (FBP) that adaptively adjusts the learning rate according to systematic errors. The prediction of cotton bollworm, for example, compares the convergence rates of the two algorithms.