回声状态网络在半自磨机功率预测中的应用

采用回声状态网络(ESN)对半自磨机功率进行了预测, 并成功应用于某现场半自磨机的功率预测中。与实时监测结果对比发现, 该算法对磨机功率预测具有较高的鲁棒性和准确性, 同时能结合选矿厂实际情况, 具有较好的实用性和广泛性。该方法不仅可用于磨矿过程磨机的自动化控制, 同时可在现场检测仪表出现故障时为现场磨机控制提供必要的参数指导。采用ESN回声状态网络对半自磨机功率进行预测可提高半自磨机磨矿效率、防止磨机涨肚、降低球耗及电耗。

Application of Echo State Network in Prediction of Power of Semi-Autogenous Mill

Echo state network (ESN) was employed to predict power of semi-autogenous mill(SAM) and had been successfully applied in an on-site practice. Compared with the actual monitoring data, ESN algorithm is robust and precise in predicting mill power. It can fit well with the actual operation state of the plant, exhibiting good applicability. Besides being applied in the automatic control of milling, this method can, substituting the malfunctioned monitor system, act as a parameter supervisor for on-site control of mill operation. Prediction of power of SAM using ESN can increase the milling efficiency, avoid bulging belly, as well as reduce the ball media consumption and electricity cost.