Abstract: The method was studied to exactly estimate the frequency of the sinusoidal signal as quickly as possible, which would satisfy the requirements for real-time processing and accuracy. The completed Rife algorithm provided by this paper is a fined frequency estimate that also uses another two spectrum pints beside the maximum spectrum point. In the same condition, it has a lower SNR and is more accurate than the Rife algorithm when the truth frequency is near those spectrum points. But it will have a great deviation when the truth frequency is away from those spectrum points. The synthesized algorithm which is also proved by this paper is another frequency estimate based on the detected maximum spectrum point. First, two spectrum points beside the maximum point are fined at half frequency step, and then the T is used which is the ratio of the new fined spectrum points to determine which algorithm will be empolyed. The modified Rife algorithm is employed as the T>1.5 and the fine completed Rife algorithm is employed as the T≤1.5. 500 simulation for various algorithms show that the synthesized algorithm not only has a good estimation accuracy, but also has a good frequency stability.

Key words: spectrum subdivision, sinusoidal frequency estimator, simulation