An Optimized Window Length ased Time-Frequency Representation for Non-Linear Frequency Modulated Signals

摘要

Optimal Window length based Lv Transform (OWLT) is proposed as an improved form of short-time Lv transform (STLVT). STLVT is a short-time time-frequency analysis method based on Lv Distribution (LVD). The performance of STLVT is mainly decided by the length of window. However, it has not been researched how to decide the window length in STLVT. As an improvement, OWLT achieves the optimization of window length in STLVT. Optimal window length is obtained by stochastic approximation method, which is a cornerstone of stochastic optimization. OWLT achieves global optimization by several times local optimizations starting from different origins. In simulations, two non-linear frequency modulated signals are analyzed. Results show that STLVT with the optimal window length can reach the highest signal energy concentration.