全尾砂絮凝沉降参数GA-SVM优化预测模型研究

为了得到经济、高效的絮凝沉降参数，建立GA_SVM预测模型进行优化选择。在优选过程中，以供砂浓度、絮凝剂单耗和絮凝剂添加浓度作为输入因子，以沉降速度作为综合输出因子，通过室内试验，建立训练、验证样本集；建立支持向量机（SVM）回归预测模型，用训练集对模型进行训练，进而以验证集预测值的均方误差作为适应度函数，通过遗传算法（GA）对SVM模型参数进行优化选择，应用优化得到的SVM模型对絮凝沉降参数进行预测、优化。以湖南某铅锌银矿为例，通过建立的GA_SVM模型对全尾砂絮凝沉降参数进行预测，优选出该矿最佳絮凝沉降参数为:供砂浓度20%-25%，絮凝剂单耗8g/t，添加浓度009%。经实验对比，该模型对絮凝沉降参数预测结果的相对误差能控制在5%左右，精确度较高，可以作为絮凝沉降参数优选的一种新思路

Study on GA_SVM optimal prediction model on flocculating sedimentation parameter of unclassified tailings

A GA_SVM model was established to optimize the flocculating sedimentation parameters .The tailings concentration , flocculant consumption and flocculant concentration were used as the input parameters and the sedi -mentation speed was confirmed to be the synthesized output parameter .Some training and validating samples were established through indoor experiment .Then, for predicting flocculating sedimentation parameters , a support vector machine （ SVM） regression model was established .The mean square error of the value was made as a fitness func-tion.Then, the model parameters were optimized through the genetic algorithm （ GA） .GA_SVM model was used in some mine , and the results showed that the best tailings concentration , flocculant consumption and flocculant concentration are 20%~25%, 10g/t and 0.09%.Comparing with the experiment results , the relative error of prediction result can be controlled at about 5%.The application indicates that this mode makes good effect , and it provides a new method to optimize the flocculating sedimentation parameters .