해저 연약지반 무한궤도 주행장치의 크리깅 근사모델 기반 강건설계

Abstract

The tracked vehicle of deep-sea manganese nodules miner travels on very cohesive soft soil of deep-sea floor. The performance of the tracked vehicle depends on both designation of control factors, namely design variables, and influence strength of uncontrollable noise factors, namely design parameters. In this paper, robust design methodology is applied to the design of the tracked vehicle in order to minimize the variability, or, variance of system performance from noise factors. The environmental and operational conditions such as shear strength of soil and vehicle travel speed are selected as noise factors in this study. The width of track, contract length and weight and position of C.O.G. for vehicle and buoyancy module are chosen as control factors. The sinkage, slip, pitch angle and torque are determined as system performance. The system performances are obtained using the single body vehicle dynamic simulation program, TRACSIM. Screening using L27 orthogonal array for control factors and noise factors is performed to minimize the number of control factors and noise factors. The kriging approximation model is performed in order to get effectively the response value for a design set using DACE (Design and Analysis of Computer Experiments) and MES (Maximum Entropy Sampling). Objective function is defined as the weighted summation for 4 kinds of mean and variance of system performance. Constraints are defined as 4 kinds of inequality constraints for summation of mean and variance of each system performance. Mean and variance are directly calculated from the kriging approximation model based on assumption on gaussian distribution of noise factors. Robust optimal solution is obtained by using nonlinear constrained algorithm. The robust solution is compared with classical optimization one.