Mechanism-constrained D-optimal design of a flexible claw for peanut shelling under biological variability

Experimental optimization of seed-peanut shelling mechanisms is limited by the low signal-to-noise ratio of biological materials, because variations in pod size, shell thickness, shell-kernel clearance, and rupture behavior can mask the effects of claw geometry. This study proposes a flexible peanut shelling claw and a mechanism-constrained D-optimal response surface optimization (MC-DORSO) framework to reduce biological-variability-induced noise while lowering shelling force and kernel damage.

A Hertzian-curved beam coupling model was used to guide the stress-concentration claw design. MC-DORSO combined mechanical feasibility filtering, D-optimal sampling, mean-normalized response modeling, and mixed discrete-continuous optimization. The relative shelling efficiency index R_μ and adaptive stepwise regression were used to reduce baseline force differences among pod-size groups.

The optimized parameters were θ = 120°, L_p  = 16 mm, γ = 45°, and H_t  = 0.6 mm. In validation tests against flat-press compression, the optimized claw reduced mean critical shelling force by 38.82%, increased the intact-kernel rate from 84.00% to 94.00%, and increased the germination rate from 92.00% to 98.00%.

The proposed MC-DORSO-based claw design reduces the load required for single-pod shell rupture while improving kernel preservation, providing a scalable optimization approach for low-damage seed-peanut shelling.