一种净化过程钴离子浓度的混合智能预测方法

Hybrid Intelligent Prediction Model of Cobalt Concentration for Purification Process

针对锌湿法冶炼净化过程的复杂性,提出了一种结合粒子群算法和案例推理方法的净化过程II段出口钴离子浓度混杂预测模型。考虑到不同时期案例所起的作用不一样,提出了一种综合加权相似函数。针对案例推理方法中属性权重选择和近邻个数的选取问题,提出了带有变异的惯性权重自适应粒子群算法优化方法,优化最近邻算法中特征权重矢量和近邻数,提高案例的检索精度。以净化过程生产数据进行实验验证和对比分析,计算结果表明改进的案例推理模型精度优于神经网络模型,模型预测结果可以作为过程信息用于净化过程的优化控制。

A hybrid intelligent prediction model combining case-based reasoning（CBR） with adaptive particle swarm optimization（PSO） was proposed for the cobalt concentration prediction of purification process in zinc hydrometallurgy. Owing to the different effect of the case in different periods, a combined weighted similarity functions was presented. Consideration of the retrieval accuracy of CBR influenced by the feature weighting vector selection and the optimal number of nearest neighbors, an adaptive PSO algorithm was proposed to optimize these parameters. The experimental verification and comparison analysis were executed using the industrial production data from purification process. The results show that the accuracy of the hybrid intelligent model is higher than the BP neural network model and the prediction results can be used as process data for the operation optimization of the purification process.