Research On Sand And Gravel Materials Level Detection In Batching Plant Based On Machine Vision

The material level of sand and gravel storage bin(hereinafter referred to as material level)is an important reference basis for the production personnel of the concrete batching plant to master the situation of remaining sand and gravel material,control the production hours and arrange the number of sales vehicles.During the concrete production process,the material level should be monitored in real-time to further control the remaining volume of sand and gravel material within a reasonable range.However,the current material level detection work mainly relies on manual measurement,which has the problems of inaccurate measurement,high labor intensity and high risk factor.The existing material level detection sensors are expensive,not easy to install,and vulnerable to the harsh environment of the silo,resulting in equipment performance damage.Therefore,the measurement of the height and volume of sand and gravel aggregate needs newer technology to meet the demand for real-time and online detection of material level,and to provide technical support for the safe production and intelligent construction of batching plants.With the development of image processing technology and artificial intelligence,the application of machine vision in industrial inspection is gradually increasing.Machine vision-based material level detection technology will improve the accuracy of material level measurement,ensure safe operation,reduce environmental interference,and reduce detection costs.In this paper,we study the sand and gravel aggregate material level height and volume measurement models based on monocular imaging and contour generation respectively,and combine machine vision,convolutional neural network and storage bin positioning measurement,etc.,to provide a new development idea for sand and gravel material level detection in batching plant.The main research contents and results of this paper are as follows:(1)The sand and gravel aggregate material level height and volume measurement system is established,and an intelligent measurement method of sand and gravel aggregate material level based on monocular imaging is proposed.Using a monocular camera,and then constructing a specific projection model between the camera and the storage bin to establish a mapping relationship between the image coordinates and actual imaging angle,and the target detection algorithm is applied to derive the positioning information of peak and valley to find out the aggregate in single-peak and single-valley states.And finally find out the height of the material level in single-peak and single-valley states.The average measurement accuracy of the heights in the two states is 95.43%.(2)The intelligent measurement method of sand and gravel aggregate volume based on monocular imaging is designed.On the basis of obtaining the height of sand and gravel aggregate by monocular imaging,the YOLOv5 model is then used to locate the height of aggregate and the diameter of the surface circle formed in the valley state,and the corresponding locating data information is also obtained.and then the relationship between this positioning data and the height,angle and volume data is analyzed respectively to realize the volume measurement of aggregate in single-peak,single-valley,double-peak and double-valley states.The average measurement accuracy of the volume in the four states is 92.63%.(3)A method for measuring the height and volume of sand and gravel aggregate levels based on contour generation is proposed.After obtaining the state of sand and gravel aggregate and the aggregate contour line,the corresponding pixel height is obtained.The volume obtained by rotating the aggregate contour line in the camera plane around a specific curve is made the pixel volume of the aggregate.The mapping relationship between the pixel height and pixel volume with the actual height and actual volume is found.And the height and volume of sand and gravel aggregate in singlepeak,single-valley,double-peak and double-valley states are then obtained,and the purpose of visualization measurement is also achieved.The average measurement accuracy of height and volume in the four states is 93.68% and 88.33%,respectively.Finally,this paper applies the above research results practically to the visualization measurement of sand and gravel material level and volume in the batching plant of Wuhan Minghua Commercial Concrete Co.,Ltd.and verifies the practicality and usability of the research content,which has a certain promotion effect on the intelligent development of the concrete industry.