In order to provide convincing artifical touch sensations, humans should be presented with high quality haptic stimuli. In the vibrotactile domain, signals are usually displayed through mechanical actuators. Current high quality actuators exhibit a high dynamic range and have the ability to display a wide range of frequencies. However, fundamentally all actuators introduce distortions into the displayed signals. These distortions are usually nonlinear with additive noise components and they can be detrimental to some vibrotactile application scenarios that require high signal playback precision. To neutralize these distortions, we propose a signal-based equalization setup with adaptive filtering. Such a setup is very general and can be applied to any actuator in a straightforward manner. We introduce a novel adaptive filter based on Volterra and bilinear filter models that is nonlinear and more robust than previous approaches. In simulations and experiments, we show that our filter model is able to consistently outperform existing adaptive filter models and equalize vibrotactile actuators efficiently.