Research On The Establishment And Optimization Of Fresh Frozen Corn Freezing Models

Quick freezing technology can reduce the temperature of corn to the freezing point in the shortest time, at the same time the original natural quality of corn can be preserved to the maximum extent. In the quick freezing process of corn, frozen time is one of the key factors to determine the quality of frozen corn. So establishing a rational and effective frozen time prediction model is an effective way to ensure the quality of frozen corn. At present, most of the domestic and foreign researchs on the frozen time prediction model are based on empirical or numerical methods, which are too rough to predict or too complicated to calculate so that can not be applied to enterprise as reference. In this paper, we carried out analysis and research on fresh corn frozen time prediction methods with fresh corn as research object and frozen time as study focus.This paper mainly contains the following contents:First, this assay introduced the corn frozen time prediction model based on traditional numerical analysis method. In the model, corn was abstracted as an object with the characteristics of a coaxial double layers column to establish the model equation. And the one dimensional unsteady heat conduction equation was calculated layer ba layer to obtain corn frozen time. The calculation process of this method was much complex.Then, the theory of artifical neural networks was introduced. On this basis, we established the fresh corn frozen time prediction model based on BP neural network, which includes a training set with 200 sets of data, a testing set with 20 sets of data, 4 input factors and 1 output factor.The results showed that it was reasonable that we used the BP neural network model to predict the fresh corn frozen time.Furthermore, we attemptted to optimize the BP neural network model with the genetic algorithm. Results showed that the error was decreased.Finally, we learned and established the fresh corn frozen time prediction model on the basis of radial basis function(RBF) neural network. Compared with BP neural network model and optimized BP neural network model, the RBF neural network model was much better. Thus, it was reasonable and practicable to establish a model based on RBF neural network for predicting fresh corn frozen time.In this paper, the data was provided by the JiLin TianJing Food Co.Ltd, and the models were solved by using MATLAB toolbox. The results showed that the prediction model based on BP neural network was more simple and accurate than the traditional numerical method. But the BP neural network model required a lot of data to support it. Although the network model optimized by GA reduced the prediction error slightly, it shortened the network training time greatly. Compared with the above models, RBF neural network model had better applicability and accuracy. The research can provide certain reference value for the enterprise to produce quick-freezing corncarry on the corn quick freezing.