Research On Examining Apple Quality Using The Mechanical Property Based On Artificial Neural Network

The apple has an important economic value, moreover the nutrition is rich and the taste is delicious. At present, because the examination and processing method are behindhand, the fruit quality is descent, and the production, processing and sale of apple have been influenced too. Therefore, the quality examination of apple is a problem that urgently to be solved.On the base of research on the mechanical properties and the physicochemical properties index sign, combing with the relation between them, the evaluation index sign of the examining apple quality with the mechanical properties index sign was established, used nerve network model to exam apple quality, the apple quality examination system was developed based on the virtual instrument and software of MATLAB. The main research conclusions are as follows:1) According to the experiment, the index sign of mechanical properties and the physicochemical properties were obtained. The relevant research and regression analysis were done to them, the result indicates that using the apple mechanical properties index sign to describe it's physicochemical properties index sign is possible.2) Estimating the physicochemical properties of apple(including hardness, moisture content, soluble solid, total acid) by the mechanical property (including the maximum of compression, the yielding force, elastic modulus) with the method of optimization algorithm of L-M (levernberg marquardt) BP neural network, an artificial neural network of storage quality properties model was built. The mechanical properties measured in the experiment as an input and the storage quality properties as an output were to train the BP neural network. The emulational results show that this neural network makes a good estimation of apple physicochemical properties through mechanical properties. When tested by five groups of Non-sample data, the error between the estimation of this mode and the measured value is below the 5% , which can meet the precision requirement of apple storage quality properties well in engineering applications .3 ) The store model of apple quality was established with the BP nerve network. The model can predict the rule of apple's mechanical properties and physicochemical properties along with the time variation, the model is usable though the testing.