Research of loss factor identification based on observer Kalman filter identification(OKID)

Reliability of statistical energy analysis(SEA) models depends on good estimates of SEA parameters,such as coupling loss factor,damping loss factor and modal density.The power flow realization method(PRM) combined with SEA model improvement(SMI) is an approach for identifying SEA parameters from experimental data,which need not to excite every subsystem sequentially.However,the pulse response or Markov parameters of power flow system and reasonable initial model are necessary for successful updating.An improved approach based on observer/kalman filter identification(OKID) for power flow model identification is presented.The Markov parameters can be computed from general experimental data of the input power and output energy,which is then used for minimal realization of the state space representation of power flow model based on eigensystem realization algorithm(ERA).Also a refined SMI method is studied,which considers the coupling information among subsystems of initial SEA model as an additional constraint.In addition,the percentage change to each loss factors which constitute the coefficients of coupling matrix is minimized during the improvement process.The improved approach presented in the paper is validated using the test simulation of an actual structure composed of five subsystems,and performs very well with the noise perturbation.Comparing with traditional SMI,the refined technology is more flexible about the initial SEA model and the updated parameters maintain more physical sense.The identification methodology presented can efficiently improve the experimental identification precision of damping loss factor and coupling loss factor,extend the engineering application of experimental SEA parameters identification.