QAPSO-BP算法及其在水电机组振动故障诊断中的应用  

QAPSO-BP algorithm and its application in vibration fault diagnosis for a hydroelectric generating unit

作　　者：程加堂[1] 段志梅[1] 熊燕[1] 

机构地区：[1]红河学院工学院,云南蒙自661199

出　　处：《振动与冲击》2015年第23期177-181,201,共6页Journal of Vibration and Shock

基　　金：云南省教育厅科学研究基金项目(2012Y450)

摘　　要：针对水电机组振动故障耦合因素多、故障模式复杂等问题,提出了一种基于量子自适应粒子群优化BP神经网络(QAPSO-BP)的故障诊断模型。在QAPSO-BP算法中,利用量子计算中的叠加态特性和概率表达特性,增加了种群的多样性;根据各粒子的位置与速度信息,实现惯性因子的自适应调节;为避免陷入局部最优,在算法中加入变异操作;并以此来训练BP神经网络,实现网络的参数优化,进而构建了机组的振动故障诊断模型。仿真实例表明,与粒子群优化BP网络(PSO-BP)法和BP网络法相比,该算法具有较高的诊断准确度,适用于水电机组振动故障的模式识别。Aiming at problems of vibration fault with many coupling factors and complex fault modes of a hydroelectric generating unit,a method of quantum adaptive particle swarm optimized BP neural network（ QAPSO-BP）was proposed. In this algorithm,The characteristics of superposition state and probability expression in the quantum computing were adopted to increase the diversity of population. The position and velocity information of each particle was used to adjust inertia factor adaptively. To avoid falling into local optimum,mutation process was added in the approach.Afterwards,the BP neural network was trained with QAPSO to achieve the optimization of its parameters,then the vibration fault diagnosis model of the unit was established. The simulation showed that the diagnostic accuracy of the QAPSO-BP algorithm is higher than those of the particle swarm optimized BP network（ PSO-BP） and the BP neural network,and it is suitable for fault modes recognition of hydroelectric generator units.

关 键 词：BP神经网络 量子自适应粒子群优化算法 水电机组 振动 故障诊断 

分 类 号：TM312[电气工程—电机]