Objective: In order to solve the difficulty of multi-gas dynamic monitoring and early warning of fruit spoilage. Methods: Based on the gas sensor, the fruit spoilage sensor detection system was developed, and the gas sensor module, data acquisition module and other modules were designed. Developed inspection software and integrated inspection system.Taking apple as the verification object, the response difference and change law of the gas sensor before apple corruption were analyzed. Results: The linear discriminant analysis, k-nearest neighbor and back-propagation artificial neural network （BP-ANN） chemometric methods were used to establish the classification model of apple before spoilage. The recognition rate of BP-ANN was the highest, the training set and prediction set were 99.53% and 99.38% respectively. Synergy interval, genetic algorithm, simulated annealing, ant colony algorithm and competitive adaptive reweighted sampling （CARS） combined with partial least square （PLS） were used to screen characteristic variables to establish the prediction model of days before corruption. The CARS showed an optimal performance in predicting the days before corruption, to achieve Rp of 0.974. Conclusion: It shows that the fruit spoilage detection based on gas sensor technology is feasible, and it provides a reference for the research and development of fruit spoilage detection system.