Sistema de medição volumétrica em fluxo de grãos de café

Abstract

Coffee is one of the world’s main commodities, with Brazil being the largest producer of this commodity. The coffee production chain is characterized by a sequence of stages, starting with production in farming, harvesting, drying, processing, storage, reprocessing, roasting and con- sumption. All phases of the production process are important in determining the quality of the final product. The stage of reprocessing the coffee beans is responsible for eliminating impu- rities and defective beans, as well as classifying the beans by size, density and color. At this stage, specific equipment cleans the coffee beans and classifies them, according to the physical characteristics of the beans. Normally, coffee is transported between reprocessing equipment in a warehouse without volumetric measurement control, making it impossible to estimate the volume of beans processed by equipment. The importance of volumetric measurement of the grain flow is also verified in the classification process, when the coffee is divided into batches according to their physical characteristics. The input quantity, related to the measurement of each batch resulting from the process, would enable the specialist to accurately assess the vo- lume necessary to form the desired blend. In this context, this work aims to develop solutions that allow the volumetric measurement of the flow of coffee beans during the transport of the product in reprocessing warehouses. The sensors identified as viable were acquired and a struc- ture for experiments was built in the laboratory to enable the continuous flow of coffee beans, allowing the application of the sensors for experimentation. Initial experiments indicated the load cell sensor and computer vision as more viable for application in a coffee bean reproces- sing environment. The experiments carried out with the prototype developed with a load cell sensor for Arduino, achieved an average relative error of 4%. Given the promising results, a future study using the industrial load cell sensor was proposed, as it guarantees greater preci- sion in measurements. The work to develop a volumetric measurement system using computer vision resulted in the development of an application for image processing and calculation of the volume in cubic centimeters of coffee beans transported on a conveyor belt. To develop the prototype, a laboratory experiment structure was built to simulate a conveyor belt. A module was adapted to the structure to capture images of the transported volume under the projection of a laser beam. Software was developed, using an innovative technique for calculating the volume of grains by processing the pixels of the captured image, allowing the measurement of the volume of grains transported by the belt. Experiments with the computer vision measure- ment prototype indicated an average relative error of around 3% in measurements with different volumes of moving grains. It is concluded from this work that the technologies studied have applicability to automate the process of measuring coffee beans transported internally in an ope- rational environment, indicating low cost for development and eliminating the need for a large structure for its implementation.