<p>This study proposes a cigarette case quality prediction method based on backpropagation neural network (BPNN), aiming to help cigarette case demander obtain product quality information in advance and support their quality control. Firstly, the mutual information algorithm is used to evaluate the correlation between input quality indices (quality data provided by suppliers) and output quality indices (quality data provided by demander), and irrelevant indices are eliminated to reduce prediction errors and simplify model complexity. Secondly, the particle swarm optimization (PSO) algorithm is adopted to optimize the hyperparameters of BPNN, and the adaptive moment estimation (Adam) algorithm is used to realize the adaptive update of network weights and biases, which effectively balances the prediction accuracy and convergence speed of the model. Finally, the proposed algorithm is compared with other mainstream regression algorithms and BPNN variants. The results show that within the acceptable training efficiency range of the application scenario, the hybrid algorithm achieves the optimal prediction accuracy and fitting ability, with SMAPE as low as 10.64% and R² as high as 0.875. The results of paired t-tests further confirm that the performance advantage of the proposed algorithm is statistically highly significant (p &lt; 0.001). Therefore, before demander conducts quality sampling inspection, the proposed prediction method can provide reliable quality prediction guidance, effectively reducing their quality control costs and improving control efficiency.</p>