Objective To address the long detection cycles, complex operation, and destructive sampling in apple freshness detection, a rapid and non-destructive detection method based on Zigbee wireless sensor network and partial least squares-discriminant analysis (PLS-DA) is proposed.Methods With Red Fuji apples as the research object, a gas sensor array acquisition system based on the Zigbee protocol is designed to monitor the concentration changes of four characteristic gases (ethanol, ethylene, oxygen, and carbon dioxide) during apple storage in real time. Least squares filtering is employed to preprocess the raw data, and the stable value, mean value, and maximum value are extracted as freshness feature parameters. A PLS-DA model is constructed to achieve rapid discrimination of apple freshness levels (fresh, sub-fresh, and rotten).Results Under indoor storage conditions of 20 ℃ and 50% relative humidity, the four gas sensors exhibit obvious response characteristics to apple volatile gases, with significant differences in gas concentrations among different freshness levels. The first three latent variables of the PLS-DA model cumulatively explains 92.47% of variance, with samples from the training set and prediction set forming clear clusters by category. The overall recognition accuracy of the model for 36 samples in the prediction set reaches 94.44%, with recognition accuracy of 100.00%, 91.67%, and 91.67% for fresh, sub-fresh, and rotten samples, respectively.Conclusion This method enables rapid and non-destructive detection of apple freshness.