基于优化BP神经网络的掘进机截割臂故障诊断

Fault Diagnosis of Cutting Arm of Roadheader Based on Optimized BPNeural Network

掘进机的截割臂是进行巷道掘进成形的最关键部件之一,对其快速准确的故障判断是智能掘进的基础和安全保障。研究了针对掘进机截割臂几种改进的BP故障诊断方法,包括传统的BP神经网络算法、基于共轭梯度优化的神经网络算法、基于Levenberg-Marquardt算法优化的BP神经网络算法和基于遗传算法(GA)优化的BP神经网络算法。通过实验仿真发现,传统BP算法难以收敛且准确性不高;基于共轭梯度优化和基于Levenberg-Marquardt优化的BP神经网络算法虽然诊断时间和精度有所提高,但并不最优;基于GA优化的BP神经网络可以在获取少量状态信号的情况下,以100%的准确性和最快速度对掘进机截割臂进行故障诊断。

Cutting arm of roadheader is one of the most important parts in the forming process of roadway tunneling.The rapid and accurate fault diagnosis of cutting arm is the basis and safety guarantee of intelligent tunneling.Studied several improved BP fault diagnosis methods for cutting arm of roadheader,including traditional BP neural network algorithm,BP neural network algorithm based on conjugate gradient optimization,BP neural network algorithm based on Levenberg-Marquardt algorithm optimization and BP neural network algorithm based on genetic algorithm(GA)optimization.Through the simulation of experimental,it is found that the traditional BP algorithm is difficult to converge and its accuracy is not high.Although the diagnosis time and accuracy of BP neural network algorithms based on conjugate gradient optimization and Levenberg-Marquardt optimization have been improved,they are not the best.The BP neural network optimized by GA can diagnose the fault of cutting arm of roadheader with 100%accuracy,fastest speed and highest precision when a few state signals are acquired.