Prediction of Diaphragm Wall Deflection in Deep Excavations Using Evolutionary Support Vector Machine Inference Model (ESIM)

摘要: Problems in deep excavations are full of uncertain, vague, and incomplete information. In most instances, successfully solving such problems depends on experts' knowledge experience. The primary object this research was to propose an “Evolutionary Support Vector Machine Inference Model (ESIM)” predict wall deformation excavation Taipei Basin. ESIM is developed based a hybrid approach that fuses support vector machines (SVM) fast messy genetic algorithm (fmGA). SVM primarily concerned with learning curve fitting; fmGA optimization. Fifty-seven monitoring database were collected data compiled from prior projects. Fifty-two 57 selected for training, leaving 5 valid cases available testing. Results show can the deflection apply contractors utilizes experience past projects new Therefore construction foundation update different stages during process, order next stage examine whether max within controlled range. results used as guidelines site safety risk management.