Abstract: When using the acoustic method for gas pipeline leak detection and location identification, we need to know the pipeline length in order to precisely detect the leakage. Here, we use a blind system-identification method to estimate the transmission performance of the two acoustic paths associated with the leak source. This method of building the cost function can solve the estimation question of the prior channel order. In order to avoid convergence to the local minima, a genetic algorithm is used to minimize the cost functions. Simulation results show that when the pipeline length is unknown, the method has high precision, small error, and is of great practical value.