Operational transfer Path analysis based on Tikhonov regularization and particle swarm optimization algorithm

Operational transfer path analysis (OTPA) address NVH (noise, vibration, and harshness) issues in various engineering but faces accuracy challenges due to noise in response data. To improve the transfer rate function’s estimation accuracy, this paper proposes a regularization parameter selection method distinct from the L-curve technique. This method formulates a regularization parameter form with parameter x based on the maximum singular value of the indicator point response matrix and the signal-to-noise ratio of the testing environment. Through the particle swarm algorithm, it searches for parameter x to find a regularization parameter that achieves a good balance between the norm of the solution and the residual norm. Simulation on a lumped-mass model and validation on a double-layer aluminum beam show the proposed algorithm achieves error rates below 20% in simulations and under 2% in experiments, outperforming the L-curve method.