In a distributed detection system, the positioning accuracy of the time difference of arrival (TDOA) method is directly restricted by the measurement accuracy of delay difference when the SNR of received signals is low. Here, an optimized TDOA localization algorithm is proposed based on multi-antenna signal combining. The combined signal is used to accurately calculate the delay difference between every two signals. Simulation results show that the proposed algorithm can reduce the positioning error by one order of magnitude compared with the traditional Gauss-Newton iterative method in the case of low SNR. A stable distributed optical fiber link is established, and further experimental results show that the positioning accuracy of the proposed algorithm can reach 1.6meters when the SNR is 1.1dB, and the positioning error is reduced by 86.8% compared with the traditional algorithm. The proposed algorithm will greatly expand the application scope of the TDOA localization method.