可檢測任意形狀與大小的智聯網、手持式色澤槍-以肉腸類的色系分選為品管示範

The color of food not only affects consumer acceptance, but also affects quality level and freshness. However, the judgment of well-trained workers will be affected by people, physiology, and background light. It is difficult to meet the needs of automation, reducing the number of people on the production line, avoiding bacterial contamination, and producing large amounts and quickly. This project introduces technologies such as visible spectroscopy and machine learning to complete the development of a prototype color sorting module, including fixtures that can change shape and design according to food items and can be thrown away/replaced for cleaning. The wavelength range of the visible spectrometer of the prototype machine is 200nm~900nm, and a total of 972 pieces of data have been completed to construct a database of spectral diagrams and color L, a, and b value data of meat sausage products, and a random forest model was used to analyze the 5-level product color data. For category prediction, the model precision, recall rate and F1 score are 0.94, 0.94, and 0.94 respectively. The accuracy of the test set reaches 93.85%, which indicates that the model is highly fitted to the training data and has good generalization ability and quite high prediction accuracy. In addition to being handheld, this prototype machine also has a color sorting module installed on the existing automation equipment (sausage cutting machine) of the partner manufacturer. When cutting smoked sausages, it can simultaneously identify, sort, and eliminate the appearance of smoked sausages. Therefore, (1) operators can create cloud work logs with the smart cloud of Internet of Things (IoT). Managers can remotely manage operator and product status history. And back-end managers can remotely create AI algorithms to link some quality control indicators. It is expected that after this case, production line testing will be introduced to improve the stability of use and design the next generation of smart manufacturing machines. (2) By improving the structure and strengthening the IoT functions, it can further strengthen technological innovation and attract existing industry users, as well as more users, who want to use it but have unmet needs. For society, more food businesses can improve product quality through this product. From an economic perspective, this testing machine can be adopted by leading industrial players as a demonstration, beneficial to market promotion and competitiveness.