Predictive maintenance poses a new way to minimize costs and downtime of machinery. The combination of sensor data, intelligent algorithms and computing power allows this new approach to monitor the current health-state of machinery and detect possible failures early on or even in advance. Previous work in this field regarding radial fans focused on aspects such as vibration and noise, whereas this paper concentrates on the influence of multiple sensor data when modeling radial fans. In a case study multiple sensors are mounted on a radial fan and the importance of their signals on damage prediction is presented. The correlation between them is analyzed and the variable impact of sensor signals for approximating the rotational speed of a healthy and a damaged radial fan is identified.