Equipment consumption prediction model based on PSO-SVR algorithm

The accuracy of equipment consumption forecasting is vital for determining the operational effectiveness of military support systems. As equipment systems evolve toward higher levels of informatization, system integration, centralization, and intelligence, historical consumption data increasingly exhibit characteristics such as limited sample availability and elevat ed complexity. To address these challenges, a predictive model utilizing Support Vector Regression (SVR), with its parameters optimized via the Particle Swarm Optimization (PSO) algorithm, is proposed in this study. This approach utilizes an intelligent optimization algorithm to calibrate the parameters of the SVR model, thereby enhancing both computational efficiency and predictive accuracy. Using consumption data for high-turnover equipment Q from a designated warehouse as a case study, the dataset is processed through both standard SVR and PSO-SVR models. Simulation results confirm that the PSO-optimized model delivers stable performance across both training and testing datasets, effectively meeting the precision requirements for military material consumption forecasting.