Optimal operation control of the raw slurry blending process using the case-based reasoning and neural network

Raw slurry blending process is a key unit in the sintering alumina industry. In this blending process, raw materials are grinded and blended in the mills to produce raw slurry. The optimal operation control objective is to make the quality indices of raw slurry into their targeted ranges. Flow rates of raw materials are the key factors that affect the quality indices of raw slurry. So, in order to realise the optimal operation control objective, right set points of flow rates must be obtained. However, due to the dynamics between quality indices and control loops of flow rates with complex natures, such as strong non-linearity, heavy coupling and difficulty of description by the accurate model, such a control objective is difficult to be realised by existing control methods. An intelligent optimal control method, which is comprised of the setting layer and the loop control layer, is proposed. In the setting layer, case-based reasoning (CBR) and neural network are adopted to obtain right set points of the control loops. In the loop control layer, the actual flow rates of raw materials follow their set points obtained from the setting layer. At last, the results of industry experiments have proven the effectiveness of the proposed method.