Market demand in the wood furniture supply chain is strongly affected by seasonal fluctuations, unexpected disturbances, and multi-factor coupling, which often limits the accuracy and stability of traditional single forecasting models. To address this issue, an AMBOP-SRC model is proposed for market demand forecasting in the wood furniture supply chain, with the aim of improving forecasting accuracy, robustness, and generalization capability.

First, the wood furniture sales data were cleaned, normalized, and aggregated at the weekly level to construct a demand time series. Then, a SARIMA model was employed to capture the linear trend and seasonal patterns of the demand series, and the initial residuals were extracted. Next, an SVR model optimized by particle swarm optimization (PSO-SVR) was used to fit the nonlinear fluctuations contained in the residual series. On this basis, an adaptive movement block optimization procedure (AMBOP) algorithm was introduced to dynamically optimize the parameters and collaborative forecasting process of the hybrid model, thereby improving search efficiency and prediction stability in a high-dimensional parameter space. Finally, ablation experiments, comparative experiments, and generalization experiments were conducted, and model performance was evaluated using indicators such as RMSE and MAE.

Ablation experiments demonstrate that the absence of any component in the AMBOP-SRC forecasting model leads to a significant decline in predictive performance. The AMBOP method not only optimizes model parameters but also effectively balances the relationship between seasonal fluctuation modeling and emergency response, achieving deep adaptation to the characteristics of the furniture industry. Performance comparison experiments show that the AMBOP-SRC forecasting model significantly outperforms classical forecasting methods in the corresponding field, effectively improving the accuracy and interpretability of wooden furniture supply chain demand forecasting. Generalization experiments indicate that the AMBOP-SRC forecasting model exhibits strong adaptability across different datasets, making it applicable to demand forecasting in key stages of the wooden furniture supply chain.

The AMBOP-SRC forecasting model is highly suitable for wooden furniture supply chain demand forecasting, featuring high prediction accuracy and excellent generalization performance. Its results provide crucial decision-making references for furniture enterprises in optimizing inventory turnover rates and dynamically adjusting production plans, thereby enhancing supply chain resilience.

To evaluate and study the vulnerability of furniture supply chain, in order to provide reference for furniture supply chain risk management decisions from the perspective of supply chain vulnerability.

Based on the new characteristics of furniture supply chain development in the context of the new era, the vulnerability of each link in the furniture supply chain was decomposed, and the Delphi method is used to screen the factors of furniture supply chain vulnerability. SPSS software wass used for reliability and validity analysis, and a furniture supply chain vulnerability evaluation index system including digital supply chain maturity is constructed. Utilizing the advantages of BP neural network in risk assessment, combined with permutation feature importance algorithm, Python was used for simulation training to construct a furniture supply chain vulnerability evaluation model based on BP neural network.

1) a furniture supply chain vulnerability evaluation index system consisting of 3 primary indicators, 9 secondary indicators, and 24 tertiary indicators was constructed; 2) The permutation feature importance algorithm was used to calculate the weights of the evaluation indicators for furniture supply chain vulnerability. Based on the weights, the evaluation indicators were ranked and it was found that the five indicators of market demand prediction, supply chain information collaboration, supply chain decision-making level, industrial structure adjustment, and product competitiveness have significant positive significance for predicting the vulnerability of furniture enterprise supply chains; 3) Through iteration and training, it was found that the furniture supply chain vulnerability evaluation model based on BP neural network has a classification prediction accuracy of 100% for 61 sets of training data, with a maximum relative error of 0.002 256%; The classification prediction accuracy for 20 sets of test sample data is 95%, with a maximum relative error of 0.5%.

The furniture supply chain vulnerability evaluation model based on BP neural network has good nonlinear mapping and learning ability, strong classification and prediction function, and can comprehensively and effectively evaluate the vulnerability of furniture supply chain.