中文摘要：

针对诺西肽发酵过程中菌体浓度的估计问题,提出了一种并联型混合建模方法。该混合模型分两部分：机理模型部分和误差补偿模型部分。利用二氧化碳释放率方程与菌体生长动力学模型,推导出了一种新的菌体生长动态模型,并以此作为混合模型的机理模型部分;利用神经网络构成误差补偿模型部分,其中该部分的辅助变量是在分析与诺西肽发酵过程对应的非结构模型的基础上,根据隐函数存在定理选取的。实验结果验证了所提方法的有效性。

英文摘要：

A parallel hybrid modeling method was presented for the estimation of biomass in Nosiheptide fermentation process. The hybrid model consists of two components： a mechanism model and an error compensation model. By using the equation of carbon dioxide evolution rate and the model of bacteria growth kinetics, a new bacteria growth dynamic model was deduced to constitute the mechanism model. The error compensation model was constituted by neural network, and based on the analysis of the unstructured model corresponding to Nosiheptide fermentation process, the secondary variables of the error compensation model were selected according to the implicit function existence theorem. The testing result shows the effectiveness of the presented approach.