Real-time optimization compensation method based on a novel two-level multi-block hybrid model for the hydrometallurgy process

Abstract

Considering the mismatch of complex process models under the influence of external environment changes and uncertain disturbance, the optimization results are not the real optimal values, and the process operating performance deviates from the optimal operation, a real-time optimization compensation method based on a two-level multi-block is proposed. Firstly, the interval plant-wide hierarchical optimization (IPHO) method is used to perform the uncertain optimization for the hydrometallurgical plant-wide process. The hierarchical optimization method is adopted, which simplifies the optimization model and realizes the hierarchical compensation operation when the process is running at the non-optimal grade. Furthermore, the interval number-based uncertain optimization method is utilized to solve the uncertain optimization problem in the presence of uncertain disturbance variables under the upper procedure-layer. Secondly, the data compensation method based on just-in-time learning (JITL) is constructed, which realizes the real-time compensation operation of the production index set-points obtained by the optimization of the procedure-layer. Thirdly, to realize the optimization compensation of the lower process layer, the real-time optimization compensation method based on self-optimization control (SOC) is proposed to compensate the manipulated variables set-points of each sub-block. Therefore, within a certain disturbance range, the operation compensation of the non-optimal operation process can be realized by performing the above-mentioned hierarchical compensation decision in real-time. Finally, an industrial simulation experiment of the hydrometallurgical production process is carried out. Compared with the traditional operation optimization and compensation methods, the proposed method can realize the self-optimization adjustment of the non-optimal operation process faster and more effectively, which can minimize the economic loss of the plant-wide production process.