非线性动力系统多阶段辨识模型、算法及应用

本文针对一类发酵过程中的发育期、生长期及稳定期的不同特性，提出一种非线性多阶段动力系统及其辨识模型。论述了该模型是一种特殊的多阶段最优控制问题。根据子控制问题性能指标下水平集的局部一致有界性及下半连续性，证明了子控制问题可控性及最优解集为非空紧集。依此性质构造优化算法，研制相应的数学软件，并应用于间歇发酵过程中的参数辨识。其数值结果表明：该多阶段模型比已有模型更能逼近实际过程，提高了模型的精度，使模型更为有效。

Multin stage Identification Model, Algorithm And Appication of Nonlinear Dynaic System

According to the different characters of lag phase,log arithmetic phase and stationary phase in a kind of fermentation process,we propose a more stage nonlinear dynamic system and its identification model.Because the level set of subcontrol's performance index is locally uniform level-bounded and lower semicontinous,we prove that the subcontrol problem can be controlled and its optimal solution set is nonempty compact.Then we construct the optimal algorithm,develop the relevant mathematic software,and apply it to the parameter identification in Batch anaerobic cultures.Its numerical result shows that this more stage model can not only approach to the practical process better than we've used before,but also promote the precision and make it more effective.