ABSTRACT

An industrial water injection hydrocyclone was adopted in the Taixi coal preparation plant to overcome the shortcoming of fine particle entrainment in the underflow and improve the grading efficiency. Parameter optimization of the hydrocyclone was critical prior to putting it into production. A particle swarm optimization optimized support vector machine (PSO-SVM) model based on uniformly designed experiments was built. It was satisfactory in accuracy as the average of the absolute of relative error was 1.73% for the training dataset, and the relative error of the test dataset was 8.43%. The trends of grading efficiency versus underflow port diameter, inlet pressure and water injection pressure were analyzed, respectively. It was found that the maximum efficiency was obtained when underflow port diameter was 50 mm, inlet pressure was 0.25MPa and water injection pressure was 0.04 MPa.

摘要

台西选煤厂采用工业注水旋流器，克服底流夹带细颗粒的缺点，提高分级效率。在投入生产之前，水力旋流器的参数优化至关重要。建立了基于均匀设计实验的粒子群优化支持向量机（PSO-SVM）模型。训练数据集的相对误差绝对平均值为 1.73%，测试数据集的相对误差为 8.43%，准确度令人满意。分别分析了分级效率与底流口直径、入口压力和注水压力的变化趋势。发现当底流口直径为 50 mm、入口压力为 0 时效率最高。