基于粒子群优化的磨矿分级过程多层数据协调

准确稳定的过程数据是选矿厂进行过程优化控制和决策管理的依据,今针对磨矿分级过程数据特点,建立了多层数据协调模型,包括总物料平衡层、粒度分布/品位层和不同粒度下的成分分析层(金属分布率层);针对模型维数较高的问题,引入粒子群优化(PSO)算法进行求解。根据不同的测量信息,可选择相应的层次进行协调,并采用从低层向高层逐层协调的方法,实现了部分非线性约束到线性约束的转化,提高了数据协调效率。将该多层模型和PSO算法用于某选矿厂磨矿分级过程实际生产数据的协调,结果表明协调后的数据更准确、更稳定,包含的信息更丰富完整。

Multi-Layer Data Reconciliation for Grinding-Classification Process Based on Partical Swarm Optimization

The accurate and stable process data are the basis for the decision-making,management and process optimal control of the ore dressing plant.Based on the data characteristic of the grinding-classification process,a multi-layer data reconciliation model was proposed,which consists of global material balance layer,particle size distribution/grade layer and chemical compositions of different particle sizes layer(the metal distribution ratio layer).The particle swarm optimization algorithm(PSO) was introduced to solve the model with high-dimension.According to the difference of the measurement information,the proper layer can be selected.The data reconciliation was conducted from the bottom layer to the top layer so that the non-linear constraints were transformed to linear constraints,which improves the efficiency of the reconciliation process.The application of the multi-layer model proposed and the PSO algorithm to adjust the actual data of the grinding-classification process obtained from an ore dressing plant shows that the adjusted values of the data are more accurate and stable,and more complete information are obtained during the reconciliation process.