Nonparametric estimation for planetary soil based on numerical simulation of penetration process

Abstract

Penetration detection is an important method for in-situ scientific exploration of planetary surfaces, and is used to indirectly detect the mechanical properties of planetary subsurface soil. In this study, four ogive-nosed penetrators with different radii were simulated to penetrate planetary soil of varying compactness under different impact velocities using a nonlinear finite element (FE) method based on HyperMesh/LS-DYNA. A nonparametric estimation method was proposed through analysis of the corresponding results. Three main characteristic parameters in the penetration process were adopted in the method: maximum deceleration am, penetration depth z, and trajectory deviation xp. Twenty parameters for identifying planetary soil properties were derived based on these three characteristic parameters. The terrain consisted of simulant soil classified non-parametrically and artificially defined as three states (low, medium, and high compactness) based on different harnesses, and five estimation criteria (slack, ideal, partially strict, relatively strict, and strict) were proposed. Three identification parameters were derived through the analysis of the evaluation indices (identification rate ηi, accurate rate ηa, and conservative rate ηc) under different estimation criteria. am2xp2 was selected as the optimal identification parameter and the superior estimation criterion is the ideal criterion. The verification results of simulation tests show that the accurate success rate and conservative success rate of the final evaluation are 60% and 80%, respectively, indicating that the nonparametric estimation method in this study can be used to estimate the mechanical properties of planetary soil effectively.