Determination of working states of the rotating cutting assembly in forage harvesters by artificial neural networks

Abstract

Abstract This work describes an algorithm which is able to determine the working states of a rotating cutting assembly automatically. The approach was validated at a self-propelled forage harvester under different environmental and harvest conditions. Data were recorded throughout different field trials near the cutting assembly using two built-in vibration sensors. The working states of the cutting assembly were divided into processing, neutral and grinding. The analysis was performed using evolutionary optimized Artificial Neural Networks. The generated models for classification are able to determine the working states robustly for this type of a rotating cutting assembly. Case-specific and sensor-specific confusion matrices are presented for performance evaluation. As a conclusion vibration data is suitable for automatic and robust classification in this context.