Abstract: It is a little difficult to recognize convolutional codes with different code rates when cepstrum, phase ambiguity and bit errors exist. A strong fault tolerance algorithm is proposed in this paper to solve this problem. This paper takes (133,171) convolutional codes for example under the QPSK modulation mode. Firstly, the transformation relationship between symbol information and bit soft information is derived. Then, check vectors under various conditions are obtained by using the check vector solution algorithm proposed in this paper. Finally, performance of check vectors is tested by using three methods which include Walsh-Hadamard transform(WHT), log-likelihood ratio (LLR) and likelihood difference (LD). Simulation results show that for (133,171) convolutional codes, we can identify different code rates, degree phase ambiguity and cepstrum effectively at low snr by using check vectors. Moreover, the algorithm reduces the analysis range from 56 to 2 or 4, has low computational complexity and it can meet the needs of the actual environment.