In order to accurately locate the fault sensor position in the structural health monitoring system, a sensor fault localization method based on the cumulative residual contribution rate is proposed. Based on the basic principle of principal component analysis, the data collected by the sensor under the vehicle load or the ground pulsation excitation are divided into the main element space and the residual space, and the fault is detected by SPE statistic. Furthermore, the residual contribution value is further deduced, and the cumulative residual contribution rate index is proposed. It improves residual contribution graph, also improves the accuracy of fault location. And sensor fault location is extended to simultaneously locate two fault sensors. The numerical simulation results show that the principal component analysis can accurately identify the four kinds of common sensor faults. The cumulative residual contribution rate not only better locates the single sensor fault, but also accurately locates the faulty position when the two sensors fail simultaneously.