基于粒子群算法的抽油机故障诊断研究

Research on Pump-jack Fault Diagnosis Method Based on Particle Swarm Optimization

提出了一种动态改变学习因子的粒子群算法,用以保证在粒子群优化算法的初始阶段,使粒子在进化初期仔细地在自身的邻域内搜索,防止粒子快速向局部最优解汇聚而错过自身邻域内可能存在的全局最优解,而在进化后期,使粒子快速、准确地收敛于全局最优解,提高算法收敛速度和精度。利用改进后的粒子群算法优化神经网络的权值和阈值,并把优化后的神经网络应用到抽油机故障检测中,结果表明用改进后粒子群算法优化的神经网络对抽油机进行故障诊断较传统BP算法更具准确性与快速性。

A new particle swarm optimization（VCPSO） based on unifying the study factor was proposed,to ensure the particles careful search in the neighborhood of its own in the earlier stage,prevent the particles fast convergence to a local optimal solution for having missed theirs own neighborhood that may exist in the global optimal solution.Particles were rapidly and accurately converged to the global optimal solution and the algorithm convergence rapidity and accuracy in the later stage was improved.The connecting weights,thresholds and structure of the neural network were optimized by the new particle swarm optimizers.The new neural network was used in pumping unit fault intelligent diagnosis system.The diagnostic results between the new VCPSO and BP algorithm were compared.The conclusion is that the network based on VCPSO has better training performance,faster convergence rate and higher accuracy.