Real-Time Optimization and Application Based on Correlation Integration with Robust Control Solution

For the traditional real-time optimization algorithm of correlation integration, when the adjustment weighting matrix of tuning variable is not set properly, the value of the objective function may diverge and cannot converge to the optimal value, a correlation integration real-time optimization method based on robust control solution is proposed. First, based on data-driven, the gradient of the system is obtained by correlation integral algorithm. Then, the state space model of the system with gradient as state is obtained by least square identification, and a robust state feedback controller with incremental constraints of tuning variables is designed. Finally, the optimal control law is obtained by solving linear matrix inequality, so that the tuning variables meet the real-time constraints in the optimization process and converge to the optimal working point quickly. The proposed method has been applied to the xylene optimization process and the industrial application results have verified the effectiveness and feasibility of the proposed method, which provides a new and effective way to solve the real-time optimization problem of complex industrial processes.