Pattern Recongnition Algorithms In The Test Of Agricultural Products Used Enose

Electronic nose is an instrument with the ability of identifying single and complex odor which involving sensor technology, electronics technology, computer, applied mathematics and other subjects. It has the nondestructive and fast characteristics and demonstrated the great potential in food quality evaluation and other fields in recent years.This article first describes the principles and structure of the electronic nose system, and then mainly studies the software of the electronic nose system for detecting agricultural products. It also carries out different recognition experiments for typical agricultural products and constructs pattern recognition algorithms based on principal component analysis and BP neural network.The sample of experiment data to typical agricultural products is preceding using principal component analysis. The results show that the system can effectively distinguish between fresh and deteriorated samples and even to distinguish the different corrupting stage; it provides a rapid and exact identification measure to distinguish the quality of agricultural products using two principal components instead of primary ten sensors. In addition, an improved BP neural network is built and trained using the physicochemical values of the samples, which results in the correlation model between sensor responses and physicochemical values. The test showed that it can achieve good predictions to unknown samples.According to the characteristics of system hardware, the running mode of system and the results of pattern recognition algorithms, we developed two sets of software for the purpose of identifying the quality of agricultural products based on the development of electronic nose system and the experiments of our research group. The electronic nose system consists of several basic modules: sensor array, gas chamber, sensor signal adjusting module, MCU module, ARM embedded systems, and computer. The electronic nose software system is written by Visual basic and visual studio language that provides a friendly interface. It can realize serial communication between host and slave device, control the experiment flow, display and save the sensor data in real time.