Abstract: To overcome the common shortcomings shared by the existing methods: weak suppression noise interference and feeble performance of time-frequency concentration, a novel time-frequency analysis method based on sparse representation was developed , which could get clear and concentrate time-frequency representation. Firstly, the unconstrained sparse representation model of FH signals was established according to the punish function theory. Then, the guideline of punish parameters were analysed theoretically and got time-frequency representation by sloving the optimization problem used approximate l0 norm finally. The simulation results show that this method is capable of getting clear time-frequency pattern.