基于BP算法和Petri网的柔性制造系统故障诊断

摘要

Aiming at the traditional Petri net can not predict precisely the appearance of fault and the lack of study ability, the BP nerve network is combined with weight fuzzy Petri net. A fault diagnosis model-BPFPN net is defined and a algorism for constructing BPFPN net model is proposed. Then BP nerve network algorism is applied to BPFPN net fault diagnosis to realize the train for the parameters. Finally, the BPFPN net model is applied to the instance of FMS on the condition of parameterψ = 5000, studying rate η =0. 05 and studying errorΔe = 0. 0002 and BPFPN net was proved to fault diagnosis.%针对传统Petri网难以精确描述故障发生的不确定性以及缺乏学习能力的缺点,将BP神经网络和加权模糊Petri网相结合,定义了一种新的能对故障进行诊断的模型——BPFPN网(A net based on BP and FPN),并提出了对BPFPN网故障诊断模型进行构造的算法,以及一种将BP神经网络算法应用于BPFPN网故障诊断模型实现对各种参数进行训练的方法；最后通过对实验参数为φ=5000,算法学习速率η=0.05,学习误差△e=0.0002的柔性制造系统加工中心故障诊断实例进行实验,在对各种参数进行学习后,能够有效地实现对故障的诊断,证明了BPFPN网是一种有效的故障诊断方法.