泡沫图像特征驱动的锑粗选加药控制策略

锑粗选工序的加药控制直接影响精选与扫选的性能.通常由人工观察泡沫手动调节药剂.这种方式,存在控制滞后、主观随意性大、易导致浮选性能不稳定甚至恶化的问题.对此,我们提出一种泡沫图像特征驱动的锑粗选加药控制策略.利用概率支持向量回归方法建立基于锑粗选关键泡沫图像特征与加药量的入矿品位估计模型;在此基础上,采用操作模式匹配方法实现加药量的预设定,快速满足入矿品位类型变化后新的控制要求;并采用基于区间II型模糊系统的加药反馈控制器减小泡沫状态与期望的偏差.工业验证结果表明,该方法能有效代替人工加药并改善了锑浮选性能.

Froth-image-features-driven control strategy for addition rates of reagents in stibium rougher flotation

Dosing control for stibium rougher flotation has a direct impact on the performance of cleaner and scavenger flotation. Conventionally, operators regulate the addition rates of reagents through observing the froth status, which is with large delay and is highly subjective causing the flotation performance unstable and even deteriorated. To deal with this problem, we propose a froth-image-features-driven control strategy. Using probabilistic support vector regression method, a feed-ore grade estimation model is built based on the key froth image features and the addition rates of reagents; then in order to satisfy the new control requirements as soon as possible when the feed-ore grade is changed, we adopt the operational pattern method to preset the addition rates of reagents; and a feedback dosage controller is developed based on interval type II fuzzy control system to minimize the deviation between the froth status and their expectation. The industrial experimental results show that the proposed control strategy can substitute operators in regulating the addition rates of reagents, and improve the flotation performance.