At present, state monitoring is a significant kind of know and master the engineering structure state at the process of construction and use. Due to the variety of the monitoring approaches and the inconsistence of the dimension of monitoring data, it is a great challenge to comprehensively analyze the monitoring data. In this study, we present an integrated information extraction method for the multi-source heterogeneous monitoring data. First, different types of original monitoring data are pre-processed and denoised based on wavelet decomposition technology. Second, the same type of monitoring data from different monitoring points are fused to one monitoring series by virtue of entropy weight method. Third, integrated monitoring information acting as time series are extracted from different types of monitoring series by virtue of principal component analysis method. At last, we use the proposed method to analyze the dynamic safety process of an irregular slab bridge under the condition of the pile foundation underpinning. The results show that the integrated monitoring formations could perfectly reflect the whole dynamic mechanical behavior of the bridge during the process of pile foundation underpinning, and the moment breaking the existing piles is the most dangerous moment in the whole process.