摘要
Emerging countries and traditional industries may not be ready for direct migration of Industry 4.0. In particular, the broiler is a major source for meat, while poultry farming in emerging countries is mainly by small and medium-sized enterprises (SMEs) or family businesses. The live broilers need to maintain the desired specification for food processing while optimizing the feeding conversion rate for revenue management. Conventionally, broiler growth monitoring and prediction rely on farmers’ experience and a small amount of data by unrigorous sampling. Although the automatic weighing system (AWS) has gradually been employed to replace manual weighing to reduce cost and casualties, existing automatic weighing systems have limited capability for weight monitoring and weight prediction of broiler future growth. To fill the gaps, this study aims to employ Industry 3.5 as a hybrid and develop a data-driven framework for weight monitoring and prediction to support smart production for poultry farming for revenue optimization. The proposed framework contains two modules. The weight monitoring integrates Gaussian Mixture Model, bootstrapping resampling, and weighted mean technique to estimate the current weight of live broilers in the farm via big data including electronic signals collected from the farms via multiple sensors and devices. The weight prediction module employs mathematical growth functions as a basis and daily feedback for adjustment to provide real-time weight prediction to support related decisions for smart production. An empirical study was conducted in Taiwan. The results have shown the practical viability of this approach. The developed solution is implemented in the broiler industry.