The mechanism of cream cheese fermentation has a complex, which makes its modelling a challenging task. Researchers have proposed different approaches to modelling the fermentation process. The developed models can be categorized into white, grey, and black box models. In this paper, we studied these models and investigated their application by using lab and industrial scale data which were obtained in the presence of disturbances. The results showed that using the states of the white box model for predicting pH is challenging because of the complexity of the cream cheese compound. Although this problem was solved in the grey box model, there were difficulties in applying both white and grey box models mainly due to the lack of online measurements of states. Unlike white and grey box models, a black box model, an ANN model, was developed based on pH data, which are measured online. Using the experimental pH data, ANN model configurations with optimal feedback and time intervals were used to predict industrial fermentation pH dynamics. The ANN model provided reliable pH predictions at both lab and industrial scales.

奶油干酪发酵的机制很复杂，这使其建模成为一项艰巨的任务。研究人员提出了不同的方法来模拟发酵过程。可以将开发的模型分为白色，灰色和黑盒模型。在本文中，我们研究了这些模型，并利用存在干扰的实验室和工业规模数据研究了它们的应用。结果表明，由于乳脂干酪化合物的复杂性，使用白盒模型的状态预测pH值具有挑战性。尽管在灰盒模型中解决了此问题，但由于缺乏在线状态测量，因此在应用白盒模型和灰盒模型时都存在困难。与白盒和灰盒模型不同，黑盒模型，人工神经网络模型，是根据在线测量的pH数据开发的。使用实验pH数据，具有最佳反馈和时间间隔的ANN模型配置用于预测工业发酵pH动力学。ANN模型可在实验室和工业规模上提供可靠的pH预测。