An improved gradient descent method with multiple variable step sizes for identifying nonlinear parameters of moving-coil loudspeakers is proposed. This method monitors parameters trends during identification and adaptively multiplies or attenuates corresponding step sizes, eliminating the need for manual adjustments. Simulations and experiments reveal that it is as accurate as the fixed step size method and more efficient and robust against initial linear parameter errors.
http://www.aes.org/e-lib/browse.cfm?elib=21026
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.
Learn more about the AES E-Library
Start a discussion about this paper!