基于STA-LSSVM的槽电压优化方法

Cell Voltage Optimization Method Based on STA-LSSVM

针对铝电解过程中,槽电压的参数调节主要采用“试探法”、过于依赖技术员、耗时长等问题,提出了一种基于状态转移算法(State Transition Algorithm,STA)的槽电压优化方法。首先,采用灰色关联度分析方法确定槽电压模型的输入变量;然后,基于改进的最小二乘支持向量机(Least Square Support Vector Machine,LSSVM)建立槽电压软测量模型;最后,采用状态转移算法对槽电压优化控制模型进行实验,获得槽电压优化值和一组优化操作参数。结果表明,建立的槽电压软测量模型具有较高的预测精度,STA算法可寻到3.8197 V的优化槽电压值,相较于优化前降低了115.8 mV,每吨铝直流电耗降低了363 kW·h,实现了槽电压的优化控制,达到了较好的节能降耗目的。

In the process of aluminum electrolysis,the parameter adjustment of the cell voltage is mainly based on the“heuristic method”,which relies too much on technicians and takes a long time.A cell voltage optimization method based on state transition algorithm(STA)is proposed.Firstly,the input variables of the cell voltage model are determined by the grey relation analysis method.Then,based on the improved least square support vector machine(LSSVM),the cell voltage soft sensing model is established.Finally,the optimal control model of cell voltage is experimented with the STA,and the optimal value of cell voltage and a set of optimal operating parameters are obtained.The results show that the established cell voltage soft sensing model has high prediction accuracy,and the STA can find the optimal cell voltage value of 3.8197 V,which is 115.8 mV lower than before optimization,and the DC power consumption per ton is reduced by 363 kW·h,which realizes the optimal control of the cell voltage and achieves better energy saving and consumption reduction.