基于改进支持向量机的石煤提钒行业清洁生产评价研究

采用遗传算法(GA)对支持向量机(SVM)进行改进,并将其应用于石煤提钒行业清洁生产评价.在系统研究石煤提钒工艺类型的基础上,根据前期已建立的石煤提钒行业清洁生产评价指标体系,提出GA改进SVM的应用思路,通过对3种工艺类型企业的现场数据采集,形成训练和测试样本,并利用GA算法确定出各类参数(惩罚参数C和核函数参数g),分别为强酸浸工艺C=2.1049,g=5.2184;弱酸浸工艺C=0.0035286,g=1.9947;水浸工艺C=0.39587,g=1.4105.GA-SVM模型测试结果表明,分类精度达到100%.通过与其他评价方法对比表明,训练好的GA-SVM方法针对小样本数据在分类精度和可操作性上都较其他方法有明显优势,实现了对石煤提钒行业清洁生产水平的定量评价.

Assessment for cleaner production of extracting vanadium from stone coal based on the support vector machine

We improved the Support Vector Machine (SVM) using Genetic Algorithm (GA) and applied it to the cleaner production (CP) assessment in the industry of vanadium extraction from stone coal in this paper. The application of improving SVM by GA was proposed based on the current assessment indicator framework of CP in the business and the systematic research on the process types of vanadium extraction from stone coal. By analyzing the acquired data collections from three different enterprises, a series of parameters of each process were determined using the GA algorithm. In the strong acid leaching process, C value was 2.1049 and g value of 5.2184. On the other hand, in the weak acid leaching process, C and g were lowered to 0.0035286 and 1.9947, respectively. In the water leaching process, C and g was 0.39587 and 1.4105, respectively. The results of GA-SVM improvement model showed that the classification accuracy could be as high as 100%. We also made comparisons with other assessment method, it was demonstrated that the well-trained GA-SVM method had significantly improved the classification accuracy and operating process of a limited number of data samples. The GA-SVM can be used to quantitatively assess the CP level in the industry of vanadium extraction from stone coal.