Strength-Driven Inverse Modelling of Agricultural waste Concrete Using Feasibility-Guided Generative Intelligence Operations

Rice husk ash and corn cob derivatives are popular in sustainable concrete systems, but the lack of dependable, performance-driven mix design procedures limits their use in structural applications. Most existing processes adjust mix proportions to desired strengths using forward prediction models or laborious trial-and-error methods. Such methods struggle with the inverse nature of real engineering issues, where compressive, tensile, and flexural strength needs are set and mix composition must obey material, durability, and sustainability constraints. The diversity of agricultural waste materials and the ill-posedness of this inverse problem make present design solutions unreliable. This study introduced a strength-driven inverse modeling method for agricultural waste concrete that handles mix design as a limited inverse job. A pipeline with five tightly coupled analytical modules is proposed. A Strength-Conditioned Mix Manifold Learning (SC-MML) module learns a continuous, constraint-aware latent representation of viable concrete mixes conditioned on goal mechanical attributes to assure initial physical plausibility. A Pareto Inverse Solver with Feasibility-Gated Diffusion (PIS-FGD) generates many candidates mix designs that meet strength, workability, and embodied carbon criteria from this manifold. A Microstructure-Consistent Forward Twin with Material Tokens (MiC-FTMT) assesses candidate mixes using rice husk ash and corn cob property descriptors to ensure hydration and porosity consistency and increase generalization across agricultural waste sources. To reduce shortcut learning and cement reliance, the Causal Robustness and Counterfactual Mix Auditing (CR-CMA) stage stress-tests mixes under controlled interventions. Finally, CL-ALV-BER refines inverse and forward models utilizing residuals from experimental input sets. The framework maintains structural performance with high objective fidelity, minimum design iterations, and cement substitutions. Accurate inverse mix designs and a systematic methodology for performance-centric, low-carbon concrete engineering enable confident use of agricultural waste materials in structural-grade applications and advance civil engineering inverse materials designs.