Predictive model for hay and silage bale weight estimation

Bale weight measurement during grassland harvesting provides the basis for various applications in the fields of precision farming, targeted feeding, logistics planning and farm management. Weight sensors have the disadvantage of being cost-intensive, difficult to position and challenging to use in motion. This paper presents the development and validation of a sensor-based system designed to estimate the weight of bales produced by balers, particularly for models without built-in weight sensors. The primary objective is to establish a predictive model that ensures standardized weight measurement across different machines. By employing data fusion techniques, the model integrates inputs from multiple sensors to enhance the accuracy of weight predictions. Leveraging machine learning, the model achieved a Mean Relative Error of 4.96%, a Mean Absolute Error (MAE) of 41.34, and a Pearson correlation coefficient of 0.96. This method is cost-effective, as it eliminates the need for additional hardware, thus reducing both equipment and maintenance costs.