基于现代信号处理的铝电解阳极效应检测定位的研究与实现

【作者】 杨军；

【导师】 邱天爽；

【作者基本信息】 大连理工大学 ， 信号与信息处理， 2005， 硕士

【摘要】 控制阳极效应是铝电解生产中的一个重要环节,目前检测阳极效应主要是通过监测槽电流、电压的变化率来实现的,并在此基础上产生了众多的阳极效应预报方法。但是电流、电压的测量受噪声影响很大,给预报工作带来了困难。J.Xue和H.A.Oye提出一种新的阳极效应检测方法。它是通过测量阳极导杆的振动来得到电解质中阳极周围气泡的状态,进而实现阳极效应的检测。该方法基于气泡产生过程与阳极性能密切联系的认识。只是,铝电解生产环境差异性很大,增加了对信号分析的难度。本文的主要的研究目标是探索适合的阳极振动信号分析的信号处理方法,来实现阳极效应的定位检测,为最终在工业生产中实现阳极效应的定位检测和预报做好理论方法和技术上的准备。 本文的具体工作如下: 首先,对阳极振动信号进行了理论分析和实验。证实分析阳极导杆振动信号,可以有效的对阳极效应进行定位检测。 其次,对阳极振动信号进行了分析比较,从谱估计、时频分析及经验模式分解等信号分析及处理方法中,得出阳极振动信号合适的分析方法。 第三,利用小波BP神经网络对阳极效应进行了智能检测及初步的预报尝试。 第四,用VC++开发平台,开发了WINDOWS界面下阳极效应定位检测软件,初步实现了集阳极振动信号的采集、分析、定位、数据管理为一体的计算机辅助诊断系统,并为最终应用工业于生产创造条件。 最后,对所做的研究进行了总结,展望了阳极效应智能预报可能遇到的问题和解决办法。更多还原

【Abstract】 It is important to control the anode effect in industry aluminum electrolysis production. Currently, the main technology to detect the anode effect focuses on monitoring the fluctuation of the slot electric current and electric voltage. Based on the technology, numerous prediction methods are proposed. But this technology suffers from noises of electric current and electric voltage and the performance of this method degrades greatly. J. Xue and H. A. Oyes put forward a new method to detect the anode effect that needs only collecting the status of gas bubbling by monitoring the vibration of anode body. Therefore, it is possible to recognize the bubbling process that is closely related to the anode performance. However, in practice, the environment of industry aluminum electrolysis on different anode body is much different. It means that such a detection for the vibration caused by the bulling is very difficult. This thesis proposes a few methods to analyze the vibration signals of anode body. These methods are able to detect and predict the position of anode body whose anode effect is happening.The main contents of this thesis are as follows:Firstly, the vibration signal of anode is obtained by experiment and analyzed theoretically. It is proved that the anode effect can be predicted and positioned by analyzing the vibration signal of poles.Secondly, the vibration signal of the anode is analyzed and compared from the spectrum techniques, time-frequency techniques and EMD etc. And a reasonable signal processing technique for predicting anode effect is obtained.Thirdly, a BP neural network is adopted to forecast the anode effect. It can be considered as a meaningful step to predict the anode effect by artificial intelligence.Fourthly, a set of software to predict the anode effect is developed under the VC++ platform. This software integrates signal sampling, analyzing, positioning and data management.At the last of this thesis, the summary and review to the research are given. The futuredevelopment of the intelligent prediction on anode effect is also presented.更多还原