基于混合策略的转炉终点锰含量预测研究

摘要：转炉终点钢水锰含量预测，对原料添加和冶炼成本节约具有重要作用。针对凭经验预估的终点锰含量值与实际值较大偏差导致的生产成本升高的问题，建立了一种基于混合策略的改进型鲸鱼优化算法(IWOA)与最小二乘向量机(LSSVM)的转炉终点锰含量预测模型，引入柯西变异提高鲸鱼优化算法(WOA)跳出局部最优的能力；借助惯性权重增强鲸鱼算法局部搜索能力和收敛精度；提出差分变异以增加鲸鱼算法在探索末期的物种多样性和降低陷入局部最优概率。实验结果表明，IWOA LSSVM锰含量预测模型不仅在全局和局部寻优以及收敛速度有较大的提升，在误差性能指标方面优势明显，且预测误差于±0.01%间的命中率为93.3%。

Prediction research of end-point Mn content of converter based on hybrid strategy

Prediction of end point manganese content in the converter could contribute significantly to raw material addition and metallurgical saving during the smelting process. To address the problem of higher manufacturing cost brought about by the large deviation between the empirically estimated end point manganese content value and the practical value, an improved whale optimization algorithm (IWOA) and least square vector machine (LSSVM) based on a hybrid strategy for predicting end point manganese content of converter was established. First, the introduction of Cauchy mutation enabled the whale optimization algorithm (WOA) to break away from the local optimum. Then, the inertial weight was used for improving the local exploitation capability and convergence precision of WOA. At last, the differential mutation was proposed to expand the diversity of species and reduce the opportunities of falling into a local extremum at the end of exploration. According to the experimental results, IWOA LSSVM prediction model of manganese content not only greatly improves the global and local search ability and convergence speed, but also shows obvious advantages in error performance indicators, and the hit rate of prediction error within ±0.01% is 93.3%.