Three Stooges and One Master

Broiler sounds can feedback their own body condition, to a certain extent. A new non-contact broiler health monitoring method based on sound detection was proposed in this project, which innovatively transforms the broiler health monitoring problem into the multi-classification problem in machine learning. First, the audio signal collection system was designed to complete the collection of broiler sound signals, preliminary signal filtering and remote signal transmission. Second, according to the comprehensive analysis on different filtering methods, the Wiener Filtering was determined as the preferred deep signal filtering method. Third, 30-dimensional sound features from four aspects were extracted and optimized, to form the data set. Then the optimal multi-classification model based on Random Forest was trained on the Azure platform for broiler health monitoring. Finally, the visualization platform was built to complete and display the health monitoring. Multiple experimental verifications show that, for an unknown fixed-duration broiler sound signal, the achieved prediction accuracy is 98.97%.

We promise and share together. Together we share information, ideas and ideas, and make decisions together to help each member learn better. We focus not only on individual goals, but also on the value-added effects of multiple members working together.