21 / 2025-03-29 14:31:00

Research on the Localization Method of Coughing Sounds in Layer Hens

sound source localization,TDOA,GCC-PHAT,SRP-PHAT,PSO

3. 智慧牧场与动物福利 > （2）智慧养殖的大数据、机器学习和物联网应用

With the growth of large-scale poultry farming, the risk of diseases affecting laying hens has increased, making the timely detection of sick laying hens is critical. Currently, the localization of abnormal laying hens mainly relies on manual inspection, which is challenging for real-time monitoring and prone to high leakage rates. In this study, we developed a three-dimensional localization method for abnormal cries in laying hen houses based on acoustic source localization technology, which could be used for early disease detection and control. First, we employed the Generalized Cross-Correlation with Phase Transform (GCC-PHAT) combined with parabolic interpolation to enhance the accuracy of Time Difference of Arrival (TDOA) delay estimation. The microphone array layout was used for initial localization to reduce the searching area. Then, the Steered Response Power with Phase Transform (SRP-PHAT) algorithm was applied to calculate the azimuth and pitch angles of the sound source, and the Particle Swarm Optimization (PSO) algorithm was used to determine the distance between the sound source and the coordinate origin, ultimately using the angle and distance information to determine the 3D coordinates. Experimental results showed that the improved GCC-PHAT algorithm with time delay estimation was more accurate, which had the average errors of 1.79° and 2.38° for azimuth and pitch angles, respectively. The average error in distance was 0.21m, and the average errors in the X, Y, and Z directions for the 3D coordinates were approximately 0.15m. This method offers an effective solution for the precise localization of coughing sounds for laying hens.