一种改进的BP算法在实际应用中的研究

该文在应用BP网络诊断电厂制粉系统故障的研究中，采用一种改进的学习算法，即对BP网络权值的优化方法上进行了改进，并经制粉系统故障诊断的仿真试验证明，这种改进的BP算法能有效地解决网络学习时易陷入局部极小值的问题，提高网络的学习速度和诊断精度。

Application of an Improved BP Algorithm in the Practice

An improved BP algorithm that reforms the method of optimizing weights for neural networks is adopted in the research of the fault diagnosis for pulverizing system. The simulation test of the fault diagnosis for the pulverizing system is carried out. The test result shows the improved BP algorithm can effectively deal with the problem that the BP algorithm easily gets into the local minimal value,thus accelerating the training rate and improving the diagnosis precision.