作　　者： ; ; ; ; (唐骞）; (黄涌波）;

机构地区： 中南大学能源科学与工程学院

出　　处： 《矿冶》 2005年第3期98-101,共4页

摘　　要： 针对铝电解槽针振信息元诊断手段的不足,提出基于小波-神经网络技术的铝电解槽针振信息元分析与诊断的新方法。获取5种针振信息元的特征波谱作为神经网络的训练样本,建立神经网络诊断系统。分析了系统软硬件结构及其特点,并在350kA预焙铝电解槽上进行实验和仿真。实验证明该系统有很高的精度,且具有很好的应用价值。 A novel method of pattern recognition and diagnosis of working conditions in aluminum reduction cells based on the wavelet-neural network is proposed according to the shortage of diagnosis way of information elements of noise in aluminum reduction cells. Five types of frequency spectrum characteristics of noise in aluminum reduction cells are gained as training sample of ANN. Diagnosis System of ANN was developed for finding information dements of noise in aluminum reduction cells. The software and hardware and their characteristics of system are analyzed. All simulated information elements of noise are emulated on 350kA prebaked aluminum reduction cells. The high precision of this novel method and good value of application are proved by the simulation results.

关 键 词： 铝电解槽 针振 小波包分析 神经网络 诊断系统

领　　域： [冶金工程] [自动化与计算机技术] [自动化与计算机技术]