基于PSO-BP与组合矩的水电机组轴心轨迹识别

发布时间：2018-01-08 02:14

本文关键词：基于PSO-BP与组合矩的水电机组轴心轨迹识别　出处：《振动.测试与诊断》2016年01期 　论文类型：期刊论文

【摘要】：引入了一种由Hu矩和仿射矩构成的组合不变矩作为水电机组轴心轨迹的反向传播(back propagation,简称BP)自识别神经网络的输入特征向量,在粒子群优化算法(particle swarm optimization,简称PSO)的基础上,融入粒子位置越界处理和全局最优位置未更新计数器,利用改进的粒子群算法求解BP网络连接权值,水电机组轴心轨迹的BP识别速度和精度得以显著提升,采用优化思想对初步识别结果进行量化分析,提取定量的轴心轨迹形状特征参数,可为水电机组故障定位提供指南。仿真实验和应用实例表明,组合不变矩的识别方法优于Hu矩或仿射矩方法,构建的PSO-BP具备较高的收敛速度和识别精度,所提出的轴心轨迹识别方法成功应用到了水电机组动不平衡故障诊断案例。
[Abstract]:The introduction of a composed Hu moments and affine moment invariant moments as the combination of back propagation of hydropower unit orbit (back propagation, referred to as BP) from the input vectors of neural networks, the particle swarm optimization algorithm (particle swarm optimization, referred to as PSO) on the basis of the integration of cross-border processing and global optimal particle position not to update the counter weights, using the improved particle swarm optimization algorithm for BP network, hydropower units orbit BP recognition speed and accuracy can be significantly improved, the optimization method for quantitative analysis of the results, the axis orbit shape feature extraction and quantitative, can provide guidance for the fault location and the simulation results show that the hydropower generating units. Application examples, combined invariant moments recognition method is better than Hu moment or affine moment method, the PSO-BP has convergence speed and high precision, the The proposed method of axis locus identification is successfully applied to the diagnosis of dynamic unbalance of a hydropower unit.

【作者单位】：华中科技大学能源与动力工程学院;
【基金】：中央高校基本科研业务费专项资金资助项目华中科技大学自主创新研究基金资助项目(0118120050)
【分类号】：TV734
【正文快照】： 引言《能源发展“十二五”规划》明确提出“应积极有序发展水电等可再生能源”,水电作为一种可再生的清洁能源,已成为我国安全、稳定、经济和清洁的现代能源产业体系的重要组成部分。水电机组在运行过程中常受到水力、机械和电气3方面耦合影响,导致机组稳定性下降,若不及时诊

【相似文献】