The accurate tracking of key points (KPs) in large-scale aerospace component manufacturing has been a core technology. Sensor measurement network system (SMNS) is defined as the unified benchmark to realize the global position of KPs for determining position and orientation of components. However, due to multiple uncertain interferences, it is difficult to ensure the accuracy and reliability of SMNS under site conditions. To address this problem, this article proposes an accurate and systematic construction method of large SMNS considering multiple uncertain errors. First, the invalid points are eliminated based on geometric similarity of common observed information. Then, a novel model parameter identification method of SMNS with subsection dynamic weights is proposed to unify observed data from individual measurement systems in the global coordinate system. Next, a multilevel uncertainty propagation model under the combined effect of model parameter uncertainty and measurement uncertainty is presented, and a weighted fusion method based on uncertainty is further developed to calculate fused point. Finally, to further improve overall accuracy of SMNS, a configuration optimization method based on genetic algorithm by minimizing overall uncertainty is proposed to determine the configurations of measurement instruments in SMNS. The influence of a series of key factors on the SMNS accuracy is analyzed in detail. Experimental results show that compared with conventional methods, the average, root mean square, and maximum of point positioning errors under site conditions are reduced by 14.55%, 17.74%, and 17.70%, respectively. The research provides good guides of accurate construction of SMNS for in situ manufacturing of large aerospace components.