Rotational Beamforming

Rotational Beamforming (RBF) is a powerful technique used to enhance acoustic maps of rotating objects. In an ideal scenario, accurate evaluation would require the microphone array to rotate at the same speed as the object being measured. However, physically rotating a microphone array is impractical. This is where RBF comes into action: instead of physically rotating the array, each microphone is virtually rotated at the same speed as the rotating object, thus enabling to achieve an acoustical map of the rotating object.

In order for this method to work, the rotational speed (revolutions per minute, RPM) of the object must be known. This is typically achieved by using external sensors, which should be set up for obtaining a synchronized measurement between the acoustical data and the RPM signal.

There are two common ways to measure RPM:

1. If the measurement setup allows, RPM data can be obtained directly from the rotating object via built-in sensors.
2. Alternatively, if option 1 is not available and the object has rotating parts, like the blades of a fan, a laser sensor can be employed. In this case, a small reflective element is placed on the moving object. Each time this reflective marker passes through the laser beam, the sensor detects it and records a pulse, allowing the system to determine the rotational speed.

Once the synchronized RPM data has been acquired, the virtual rotation can be put into execution, taking place entirely in the time domain. In NoiseImage, this is implemented by applying a time-domain filter called rotational filter, which uses the recorded RPM signal to simulate the array’s movement.

The filter works by continuously rotating the array, in which the current angular position is being tracked. At each angular step, the software virtually rotates all microphones in the array by the corresponding amount. For each of these new virtual positions, the system determines which two actual microphones are spatially closest. Then, the signal that a virtual microphone would have captured is interpolated by using two neighboring real microphone signals.

Finally, RBF finds practical applications in analyzing various rotating objects, such as wind turbines and fans. Fig. 1 demonstrates a result obtained by applying traditional beamforming to a 7-bladed fan, while Fig. 2 shows the result using RBF. The difference is striking: with RBF, the sound sources are clearly distinguishable, whereas without it, the sound sources merge, leading to an unclear and unusable result.

Details about recommendations for conducting RBF measurements can be found in the FAQ How to avoid common mistakes in Rotational Beamforming?