Research On Rancidity Prediction And Relevant Parameters’ Detection Technique During Fermentation Process Of Rice Wine

As a traditional and social alcoholic beverage, rice wine originated over three thousand years ago. In recent years, with the constant improvement of people’s living standard and the changes of people’s consume concept, the demand for rice wine is increasing. As a result, the traditional home-based production model has been abandoned because of its deficiencies like high labor demand, low efficiency and low productivity. How to fulfill automation control of the overall production process has become the primary concern of the rice wine manufacturers. Based on the under-construction large-scale rice wine enterprise’s integrated automation project, this paper focus on the resolutions of the problems found during the design of the automatic system for formation process. Specifically, construction of the prediction system for mash rancid state during the fermentation process, development of the automatic control system for the fermentation process and the design of sampling detection system are researched. The innovation points and main work are summarized below:(1)Scheme for prediction of mash rancid state using on-line detectable parameters during the fermentation process is proposed innovatively. Because of many known and unknown reasons, the mash can become rancid during the fermentation process and this would cause a great economic loss. To solve this problem, learning from the soul of soft-sensing technique, a creative approach for mash rancid state’s prediction is proposed based on the literature available, which is achieved by using on-line detectable parameters. Six parameters from different time points are selected as input characteristic variables for this prediction system, and the selection principle is elaborated in detail.(2)Suitable algorithms are selected for prediction system modeling, and the validity of the model is analyzed by MATLAB simulation, including contrastive analysis against BP neural network model. Firstly, based on the proposed prediction scheme and the analysis of advantages and disadvantages of different algorithms, Adaptive Network-based Fuzzy Inference system together with subtractive clustering algorithm is built as the prediction system. Then, MATLAB is used to simulate the model and verify its validity. Through many simulation experiments, the best parameters are selected, that is, the best fuzzy logistic inference system structure is constructed. Finally, the simulation results of the proposed model and BP neural network model are compared, and it shows that the model proposed in this paper has better prediction performance.(3)The operation platform for parameter’s sampling and detection is invented to solve the problems of parameter’s detection process in fermentation section. The whole automatic production system for rice wine factory is designed and introduced at first. Also, the hardware system of fermentation section is constructed and described. During the construction phase, many problems of parameter’s detection process are found, for example, the detection equipments are too expensive, the demand for these equipments is quite huge, and these equipments are easy to be damaged. To solve these problems, the parameter’s sampling and detection platform system is invented, and the design principle together with the operational procedure are elaborated in detail. The invented platform system improves the utilization rate of the detection facilities greatly as well as reduces input and maintenance costs of the detection system. Thus, this invention has a good prospect of application in rice wine industry.