Study On Intelligent Cleaning Method For Green Beans Agricultural Products

With the development of economy and society, most of consumers pay moreattention on food safety problem, and then agricultural product quality detectionbecomes the focus. In the view of safety detection technology of green beansagricultural products, this paper uses machine vision, PLC automation andartificial intelligence technology to nondestructively detect the green beans withmildew and plant diseases and insect pests by taking green soy bean as anexample, and an online cleaning way is presented. To ensure the detectionequipment be operated stably and to acquire distinct product images, somedeformation and vibration analysis of the main frame and conveyor belt arecarried out.The main contents are as follows:1) This paper introduces the intelligent cleaning system for green beansagricultural products, analyzes the key mechanical components function and itsworking process, and presents the structure design scheme and workflow of themachine vision system, automatic system and the control system.2) The image preprocessing algorithm is studied for the green beans agriculturalproducts, and a preprocessing scheme is designed with the methods of imagegray-scale transformation, image geometry transform, image enhancement, colorspace transformation, morphological processing and image segmentation. Thescheme validity is verified by experiment. PC program software of the imagepreprocessing for green beans is developed.3) Product recognition and processing method is analyzed with BP-NeuralNetwork, its PC application software is developed. A three layer BP-NeuralNetwork is designed based on the system requirements,17image’s featureparameters are calculated for BP-Neural Network training, testing andrecognizing. Some practical factors are considered, such as the speed ofconveyor belt, the time of image processing and image recognition and the traveldistance etc. And then the control command is sent to PLC to accurately clear theproduct with those defects.4) This paper analyzes the deformation and vibration of the main body frame andthe conveyor belt. By including practical running external loads, the stress and deformation are discussed to test the mechanical strength and stability of mainbody frame. The nonlinear stress and deformation analysis are carried out on theconveyor belt, and the design of the roller is improved to acquire distinct andaccuracy product images. The mode analysis is carried out on the main bodyframe, natural frequency and vibration modes of main body frame are calculated.The results shows that the exciting frequency of excitation sources is far from thenatural frequency, the resonance cannot take place in current system.As a conclusion, the intelligent cleaning method for green beans agriculturalproducts in this study can obtain product feature parameters accurately andrecognize it automatically, the design of main body frame is reasonable, theconveyor belt runs stably, and product images are acquired distinctly to removethe product with those defects quickly and accurately. This work is valuable andmeaningful to accelerate and promote the development and application ofmachine vision technology in the automation progress of agricultural production.