To improve the accuracy of Ultra Wide Band (UWB) localization in Non-Line-of-Sight (NLOS) scenarios, a NLOS recognition and localization algorithm based on credibility is proposed. This algorithm utilizes UWB real-time Channel Impulse Response (CIR) features and ranging values to identify Line-of-Sight (LOS) or NLOS through one-dimensional convolution neural network, and outputs the probability of LOS or NLOS. Then this probability is used to construct credibility. Based on credibility, base station selection and improved positioning algorithms are carried out. Weighted Least Squares and Taylor (WLS-Taylor) fusion algorithm based on credibility is designed. Static and dynamic measured data in various scenarios is collected to verify the performance of the proposed algorithm. The experimental results show that the proposed algorithm can effectively suppress the influence of NLOS on positioning results, and the average positioning error is less than 10 cm in NLOS environment. In environments with relatively severe NLOS, the positioning error of the proposed algorithm is reduced by 76.94 cm compared to the WLS algorithm based on distance weighting.