基于支持向量机插值的铜电解液成分检测

在电解铜生产过程中,由于检测设备的限制,只能通过每天化验一次得到铜离子和酸根离子含量的检测数据.这些有限数据不足以反应铜电解液成分的变化规律,而基于传统插值法的插入值和现场实际值差距又很大.为解决这一问题,提出了与影响因素相关的软测量模型,并实现对数据的有效预测.鉴于支持向量机在建立软测量模型方面的优势,引入支持向量机对线性插值方法进行改进,并设计权重确定方法,定量推导出软测量估计值和传统线性插入值的权重.软测量建模方法既保证了数据在时间上的连续性,又保证了计算的插入值接近实际值.仿真结果表明了其有效性.

Detection Based on Support Vector Machine Interpolation for Copper Electrolytic Composition

In the process of electrolytic copper production the contents of copper ions and acid ions can only be detected once each day through tests,and they are inadequate to showing the regularity of variation of copper electrolyte composition.It was found that the values for conventional interpolation often differ greatly from the values obtained actually in site.To solve the problem,a soft measurement model is advanced considering the influencing factors on measurement with relevant data forecasted efficiently.The support vector machine is introduced to improve linear interpolation because of its advantages in developing the soft measurement model.And a method is given to determine the weights for the values estimated by the soft measurement and values for conventional linear interpolation.The approach proposed ensures not only the time continuity of data but also the values for interpolation to approximate to actual values.Simulation results showed its effectiveness.