A modified structure for multi-resolution analysis of frequency domain self-tuning random vibration control

Many devices connected with both commercial and military applications have to undergo a range of environmental testing to ensure that they correctly operate over a wide range of operating conditions. Moving coil electrodynamic actuators, "shakers", have become widely used in vibration testing systems as a means of applying mechanical motion to a test object. The device is fixed to the table of the shaker and an accelerometer mounted on the device or table is used for feedback purposes in the control system. The usual tests carried out by the shaker are random vibration, shock and swept sine wave. In a random vibration test the objective is that the specimen under test should follow a desired power spectrum. The paper proposes multi resolution analysis for a random vibration control algorithm. The method of frequency domain self-tuning is briefly reviewed and a full block diagram of the random vibration controller is described. The multi resolution analysis for the random vibration controller is introduced. The key idea is to split the frequency spectrum into several octaves and process each octave with a different resolution. A filter bank structure is used for multi-resolution analysis implementation. It is shown that because the spectrum of the signal is changed in the control algorithm, leakage is introduced between the octaves. The structure of the filter bank is modified, thereby removing the problem of leakage.