The present application relates to the technical field of
deformation monitoring, in particular to a time-space self-learning
early warning system for foundation
settlement risk in highway construction process, which comprises an
information processing module, a weight construction module, a
tensor analysis module, a risk calculation module and an early warning release module. In the present application, the spatial position vector and the
Gaussian kernel function are used to evaluate the distribution density and generate the coverage weight, which eliminates the local deviation caused by uneven distribution and ensures the balanced expression of regional settlement characteristics. The
lag interval is deduced by using the soil consolidation theory and the deformation direction consistency is screened, the
environmental noise is eliminated to lock the effective cumulative settlement caused by specific process, the prediction model with self-adaptive ability is constructed by combining the grey
system theory, the dynamic growth rate of the settlement cumulative value is tracked in real time, the change from static control to dynamic trend warning is realized, and the sensitivity of capturing the precursor of foundation
instability is improved, which provides a scientific basis for construction process adjustment.