This paper focuses on the data analysis method of bridge quality monitoring system under the framework of edge computing, and proposes an intelligent algorithm to support the quality assessment system. Based on the EDA method to assess data integrity, a double similarity metric scheme is designed to quantify data accuracy. A lightweight deployment scheme based on RKNN model is constructed to optimize the reasoning efficiency at the edge end by combining the model quantization technique. The validation of engineering examples shows that the change rule of the 2 metrics, histogram cosine and box-and-line diagram normal value percentage, has high consistency, and in the case of a sample capacity of 2000 and significance levels of α=0.05 and 0.01, the change rule of the cosine similarity metrics is in line with the a priori data quality judgment, and the detection result of the box-and-line diagram is roughly in line with the a priori fact. In 1000 calculations, the prediction accuracy of the RKNN model ranges from 78% to 95%, and the average calculation accuracy is higher than that of the AD and ND models. Under 10% random number share, the average accuracy of RKNN model is as high as 82.3%, exceeding 6.75% and 7.22% of AD and ND models. The research results provide technical support for bridge quality control in the whole life cycle.