Research on Spare Parts Demand Forecasting for Mechanical and Electrical Equipment Based on Preventive Maintenance

This study investigates the problem of forecasting spare parts demand for mechanical and electrical equipment using preventive maintenance. It proposes a forecasting approach that combines Principal Component Analysis (PCA) with Support Vector Regression (SVR), referred to as the PCA-SVR method. The primary goal of this research is to enhance the accuracy of spare parts demand forecasting to mitigate potential economic losses resulting from equipment failures. Initially, PCA is employed to reduce the dimensionality of various factors influencing spare parts demand, identifying crucial factors and resolving multicollinearity issues among variables. Subsequently, these key factors are utilized as input variables for the SVR model, which capitalizes on SVR’s nonlinear mapping capabilities to conduct linear regression predictions in a high-dimensional feature space. Through a case study involving a specific type of automotive engine and actual maintenance data, the efficacy and precision of the PCA-SVR model in spare parts demand forecasting are validated through data standardization, PCA dimensionality reduction, SVR model establishment, and ten-fold cross-validation ${ }^{[1]}$. The findings demonstrate that the PCA-SVR method substantially enhances forecasting accuracy and generalization capacity, offering a scientific foundation for organizational spare parts management decisions ${ }^{[1]}$. This approach exhibits extensive application potential and practical significance in the field ${ }^{[1]}$.