化工网络中的高性能微分博弈数值优化算法

High-performance Differential Game Numerical Optimization Algorithm in Chemical Networks

针对化工网络中大规模、非线性动态优化问题求解复杂度高、收敛难度大以及求解精度低等问题,在原始微分博弈数值优化算法的基础上开发了一套高性能的优化求解策略。该策略注重优化求解的初值生成,从而保证优化求解的大范围收敛;同时该策略还提出优化求解精确性提升算法,在提升求解精度的同时保证了优化结果的最优性。最后采用一个典型的化工网络作为仿真案例,验证了高性能优化求解策略的有效性。

Focusing on problem of high complexity,difficult convergence and low accuracy for large-scale,nonlinear dynamic optimization problems in chemical networks,a high-performance optimization strategy is developed based on the original differential game numerical optimization algorithm.The strategy focuses on the initial value generation to ensure the large-scale convergence of the optimization solution.Meanwhile,the strategy also proposes an algorithm to improve the accuracy,and the optimality of the optimization results is guaranteed under increasing the solution accuracy.Finally,the validity of the high-performance optimization strategy is verified with utilizing a typical chemical network as a simulation case.