Research On Optimization And Control Technology Of Green Tea Withering Proces

Green tea,as one of the unique tea products in China,accounts for about 80% of the world’s total green tea trade.As the first key process in the green tea processing,the tea water-removin can release the moisture in the leaves and inactivate the activity of enzyme by baking at high temperature.However,the traditional method relies on subjective experience heavily,which results in uneven quality of tea.Therefore,this paper takes green tea from the Qingzhen Hongfeng Mountain Yun Tea Farm Co.,Ltd as the research object.The effectiveness of the proposed method is verified from three aspects: material,process and control.The main research contents are as follows:(1)Firstly,the green tea image data set(HF-G3)is constructed.Then,a multiscale convolutional block attention module(MCBAM)is proposed and combined with multiscale depth shortcut(MDS),a lightweight tea green classification model(Shuffle Net V2 0.5xSMAU)is constructed.Additionally,a model training method combining double transfer and knowledge distillation is proposed,which can further improve the comprehensive performance.(2)Firstly,the process and quality data of tea water-removin are collected through orthogonal experiment.And data augmentation is implemented by applying fluctuations to introduce virtual samples.Then,based on the artificial neural network technology,the "process-quality" network model of green tea water-removin is built based on the existing data after completing the structural design,and the coupling relationship is established.Finally,an improved beluga algorithm(WEOA)is proposed for global iterative optimization to obtain the best combination of process parameters.The experiment proved that this combination of process parameters can make the error of the water content of the tea leaves after the water-removin within the acceptable range,which not only improves the quality of single processing,but also has guiding significance for large-scale production.(3)Firstly,an adaptive fuzzy PID controller is designed for the tea cylinder waterremovin machine.Then,an improved sparrow algorithm(CSDFSOA)is proposed to gain the quantization and the scale factor of the controller,and the construction of tea cylinder water-removin machine temperature control system(CSDFSOA-TCWMFPID)is established.The advantages of this control method in steady-state performance,dynamic response speed and anti-interference ability are proved by simulation experiments.Finally,after the completion of hardware design and selection,a prototype temperature control system is built,which realizes the precise control of the target temperature of the tea waterremovin process and significantly improves the finishing quality.