| With the increasing of fruit market in China, the quality of fruit has aroused more and more attention of consumers. In order to seize the opportunitie of time in fruit market, some fruit sellers accelerate the ripening of immature fruit. The nutritional value of ripening fruit is much lower than the normal mature, or even produce harmful substances to people’s health,so it is very important to identify the ripening fruit. With the development of sensor and testing technology, the research and application of multi-sensor fusion technology are forming hotspot. Mult-sensor fusion is the method that processes the data of different sensors which are on multiple levels and multiaspect, which result in a new meaningful information that is unable to get by any single sensor. Compared with the method which is only using a single sensor, it can obtain more accurate and reliable conclusions, so as to reduce the possible errors in the information processing. Therefore, the purpose of this article is to design a detection system based on color information and odor information and identify ripening fruit by the multi-sensor fusion technology.The detection system based on color and odor which is of multisensor fusion was designed for testing the ripening fruits in this paper, which consists of hardware part and software part. The testing system was designed to detect the color and smell information of test objects. The hardware part consists of two modules which are odor information collection module and color information collection module. Smell information acquisition module consists of closed box, gas sensor array, PCI1716data acquisition card, personal computer; color information acquisition module consists of STC89C52MCU, TCS230color sensor, RS232cable and personal computer. The software system consists of LABVIEW and MATLAB. LABVIEW is a interactive interface software; MATLAB is mainly used to analyze the data. The smell information and color information were transmitted to compute respectively through the data acquisition card and RS232cable. The JinShuai apples which produced in zhaoTong city were chosen as experiment object in this paper. The experiment objects were divided into three groups, which are immature JinShuai apples, ripening JinShuai apples and normal mature JinShuai apples.The color information and odor information of three group subjects were acquired by the designed hardware. The data which has the large deviation from the mean was removed. In order to reduce the workload of calculation, the information of six gas sensor were analyzed by principal component analytical method. The first principal component can on behalf of the information of all six gas sensors after analysis, so take the the information of the first principal component as total smell information. The pattern recognition was taken respectively by two methods which are fuzzy theory and RBF neural network. The correct recognition rate of fuzzy theory is92.63%and the correct recognition rate of RBF network is93.3%, which indicates that the two methods both can be used to identify the ripening JinShuai apples.The results show that the designed system in this paper can be used to identify ripening JinShuai apples, which achieves the goal of the design and application. |
