Improving supply chain resilience is crucial to the long-term success and sustainable development of enterprises. This paper designs the supply chain toughness evaluation index system from the actual situation, and constructs the supply chain toughness evaluation model based on the material element topable theory. Quantify the supply chain toughness index, realize the dynamic evaluation of toughness level, and mine the optimization direction. Dynamic selective integrated learning method is proposed to carry out multi-supply chain entity adaptive negotiation optimization, targeting to improve the toughness level of enterprise material supply chain. The study shows that there are 5 level I toughness indicators with correlation > 0 and average score < 80, which are concentrated in 2 level I indicators, namely, evolution capability and efficiency capability, and need to be optimized accordingly. The ratings of 4 first-level indicators, namely, readiness ability, efficiency ability, adaptive ability, and evolutionary ability, are decreasing according to level Ⅳ – level Ⅰ, and the correlation degrees are 0.758, 0.245, 0.119, and 0.145, respectively. The negotiation optimization of the toughness of the evolutionary ability is realized by using the integrated learning algorithm, and the agreement is reached after 6 times of negotiation.
