Wind turbine condition monitoring is essential task in the process of maintaining machine operation at the optimal level. It is related to ensuring the profitability of investment and the provision of security in the environment of the turbine. However, operational conditions of turbine associated with non-stationary nature of the stimulus which is the wind, impede the correct diagnosis of the machine. In addition, a multitude of parameters adversely affects the clarity of predictions and setting alarm thresholds. In the article, the authors evaluate the impact of generator output power and rotational speed on selected vibration-based feature value. The study was performed for wind turbine planetary gearbox during fault development of the ring. It was possible due to historical data consisting peak-to-peak (P2P) values together with corresponding values of rotational speed and generator output power. For the purpose of the experiment the method that bases on calculation of arithmetic mean of the data in the segments corresponding to the chosen ranges of both rotational speed and generator output power is presented. Results are given in the form of three-dimensional charts, which allow assessing the impact of parameters on the studied feature. The paper shows that for machinery operating under varying regime proposed representation might serve as a valuable method for fault detection. Additionally, authors highlight the importance of analysis of vibration-based features as a function of two variables (rotational speed and power/load).