Design And Application Research Of Outlet Pressure Control System For Rice Mills

Rice is one of the most widely distributed crops in the world and is processed to produce edible rice.Rice processing is an important way for rice to become edible polished rice,and the crushing loss caused by the process of rice crushing is the main reason that seriously hinders the quality of rice milling and seriously restricts the development of rice milling industry.Therefore,reducing rice grain damage in the whitening process is an important problem that general manufacturers urgently need to solve in the research of rice processing.Taking the rice mill in a rice milling production line as the research object,using Siemens PLC control system,based on Python-PLC information and communication technology,Python-Opencv image processing and image recognition technology,an experimental platform that can be used to identify the rice crushing rate and milling degree during the processing of rice production line is finally built,and the outlet pressure of the rice mill can be controlled.The main research contents of this paper are as follows:(1)Overall design of the rice mill experimental platform.The design objective of the outlet pressure control system of rice mill is proposed,and the core goal of the experimental platform of outlet pressure control system of rice mill is to reduce the broken rice rate and reduce the labor cost of the rice milling part of the production line.The demand analysis,platform control analysis and structural design analysis of the rice mill experimental platform were carried out,and the structure was designed as sampling device,linked particle separation device,rice grain detection device and outlet pressure regulating device.Different schemes were designed for the sampling device,the linked particle separation device and the outlet pressure adjustment device,and finally the electric actuator-driven mobile sampling tank was selected as the sampling device;Select rollers with different gaps to separate the stacked rice grains by tiling;Stepper motor is selected for outlet pressure regulation.(2)Simulation based on the bulk somal simulation software EDEM to achieve structural optimization.It mainly studies the size optimization of the core structure of the experimental platform,and simulates the movement and distribution of brown rice in the experimental platform.Determine the influencing factors and levels of the test through field measurements and 3D modeling.Finally,the three-dimensional software is used to draw the rice grain model,import it into EDEM,use the spherical particles to simulate the shape of the rice grain through stacking,and add the shear modulus and Poisson’s ratio of the rice grain between the ball and the ball through reference,and determine the contact parameters between the rice grain and the contact surface between the rice grain and the experimental platform.Simulate the real collision between rice grains and experimental platforms.The distribution of rice grains from the outlet to the visual inspection is simulated.The simulation results show that 38° is the slope of the sampling tank,the gap between the third roller and the conveyor belt is 2.5mm,and the belt speed of the conveyor belt is 0.5m/s.Based on this result,an experimental platform was built.(3)Application research based on rice mill experimental platform,Siemens PLC and Python.The rice mill experimental platform takes rice grains intermittently from the outlet of the rice mill,and then separates the linked grains,separates the rice grains from the stacked state,becomes a tiled and dispersed state,and spreads into randomly distributed rice particles,which is convenient for image acquisition of rice grains.Python and Opencv were then used for image processing,filtering,morphological processing and image segmentation of the rice grain image.Thus,in the case of rice with different moisture content: the larger the angle of the baffle of the rice mill outlet,the lower the broken rice rate,the greater the mean,contrast and entropy value,the more obvious the texture characteristics of the rice grain,the lower the milling degree,and the closer the color of the rice grain to the color of brown rice.(4)Based on the traditional rice grain detection to identify the broken rice rate and milling degree of rice particles,based on instance segmentation and deep learning,Labelme annotation is used to create indica rice broken rice,and then the data is enhanced to obtain a dataset with broken rice characteristics,and then Mask R-CNN and Solov2 are used to train with Resnet50 and Resnet101 residual networks,so as to obtain the loss function decline curve and determine the best performance algorithm for this dataset.and validate the trained model.The results show that from the average accuracy of semantic segmentation,it can be seen that the accuracy of Resnet101 in Solov2 is higher than that of Resnet50 on average,about 16%.Mask R-CNN’s algorithm is better than Solov2＿r101 in Resnet50 broken rice detection,but the most accurate is Mask R-CNN＿r101.And Mask R-CNN＿r101 is also the algorithm closest to 0 in the loss function graph.