Noise Reduction Performance And Stability Analysis Of Decentralized Active Noise Barrier

Active noise barriers apply active control on traditional noise barriers to improve their insertion loss at low frequencies.With active noise barriers that have many control channels,although using a centralized feedforward system has good performance and high stability,the system complexity and cost are high.A decentralized system with many independent units or a cluster of centralized subsystems can reduce system complexity and is more practical.In this thesis,the noise reduction performance and stability of decentralized systems are studied.First,an analytical model of a decentralized active noise barrier system is established based on the Mac Donald solution.The effects of the second source interval,the error sensor position,the channel number and the frequency on the noise reduction performance and stability of the system are studied based on the model,the relationship between the noise reduction performance and stability of the system is discussed,and the stabilities of the decentralized system and the cluster system are compared.Second,the eigenvalue distribution of a matrix related to the secondary path transfer functions of the decentralized active sound barrier system is analyzed.An optimization method is proposed to adjust the eigenvalue phase to improve the system stability so that an unstable decentralized system can be changed to stable without scarifying the noise reduction performance.The feasibility of the proposed method is verified by numerical simulations,and the convergence performance of the system after the eigenvalue phase adjustment is analyzed.Finally,the experiment results are presented to verify the noise reduction performance of the decentralized active noise barrier system and the effectiveness of the proposed phase adjustment method for improving the stability of a linear distributed decentralized system.With a 24-channel decentralized active noise barrier system built in a factory,an average insertion loss of 3.3 d BA was obtained for noise below 500 Hz.The experimental results with a 6-channel linear distributed decentralized system in an audition room show that the original unstable decentralized system converged steadily after phase adjustment and a noise reduction of more than 30 d B at the error sensors was achieved.