Аннотация:The penicillin fermentation process is a fed-batch system togenerate industrial-scale penicillin for antibiotic production. Any fault inthe fermentation tank can lead to low-quality penicillin products, whichmay cause a severe impact on final antibiotic production. In this paper, wehave developed a Gated Recurrent Unit-based Autoencoder deep learningmodel to detect faults in the batch data of the penicillin fermentation process. In particular, we have used the data shuffling strategy to minimizedistribution discrepancy from different batches generated under variouscontrolling conditions for training the deep learning model. We have alsocompared the model with the Feedforward Autoencoder and Long shortterm memory Autoencoder model for fault detection. Experimental resultsshow that our model trained on shuffled data from different batches outperformed the Feedforward and Long short-term memory Autoencoder modelwith an avergae fault detection rate of 94.74%.