Identification of Foundation Pit Construction Safety using Vector Batch Normalization based Feedforward Neural Network

Foundation pit construction safety denotes the measures and implemented to ensure a safety of personnel, surrounding environment through excavation, equipment and construction of foundation pits. However, the construction is difficult because of inherent risk associated with deep excavations like soil instability, potential collapses and water infiltration. Address these challenges are important for effectiveness of foundation pit construction safety systems and improve the resilience. The research proposed the Vector Batch Normalization based Feedforward Neural Network (VBN-FFNN) method for foundation pit construction safety which reduce the interior covariate shifts, faster during training and allowed to network converge. The proposed method ensures the exacts and data driven method for recognizing potential hazards of foundation pit construction by combination of different risk indicators. The experimental results represents that the proposed VBN-FFNN method achieves accuracy of 0.887 compared with other existing methods such as Deep Neural Network (DNN), Back Propagation Neural Network (BPNN) and FFNN.