Even for the same brand of beer, beer quality may differ in different production batches. It is important to propose a fast and effective beer quality inspection technology to control the production quality of beer. In this paper, a smooth discriminant analysis (SDA) method combined with an electronic nose (e-nose) system is proposed to identify beer gas information in different production batches. A multi-pattern recognition method is combined to classify the gas information. Firstly, using the PEN3 e-nose system, different batches of beer gas information are obtained. Secondly, an SDA method is proposed, which strengthens the linearization between gas features and improves the processing effect of gas features. Thirdly, multi-pattern recognition methods are applied and combined with multiple feature dimensionality reduction methods to demonstrate the effectiveness of SDA. The results show that SDA achieves effective dimensionality reduction of different batches of beer gas features and obtains the best classification performance with the random forest (RF), with an accuracy of 97.70%, precision of 98.47%, and recall of 98.23%, thus achieving beer gas identification from different production batches.