Large-scale underwater subsidence well bottom face cutting edge counter-force prediction and construction risk identification early warning method

The invention discloses a large-scale underwater sinking well bottom face cutting edge counter-force prediction and construction risk identification early warning method. The method comprises the steps that open caisson bottom face sensor data is obtained; by taking each sensor as a unit, performing smooth denoising on original data by adopting a Kalman filtering method to ensure the accuracy of model prediction; establishing a VAR vector autoregression model, and determining an optimal input time step parameter and an optimal output time step parameter by changing input and output step parameters of the VAR vector autoregression model; the open caisson bottom face cutting edge counter-force is predicted through a sliding window prediction method based on a VAR vector autoregression model, when a prediction result exceeds a preset threshold value, it is considered that a certain construction risk exists at the moment, early warning can be triggered in advance, early warning grades are divided according to dif