This study utilizes the Sparse Identification of Nonlinear Dynamical Systems (SINDy) algorithm to optimize the nonlinear dynamic model of the economic system based on the generalized Lotka-Volterra model. The optimized model is applied to explore the complex dynamic relationships among four variables: industrial added value, financial added value, total exports, and total imports in China, including linear, interactive, and higher-order influence relationships. Compared to the traditional Lotka-Volterra model, the model optimized by the sparse identification algorithm demonstrates higher accuracy in fitting and short-term forecasting. Additionally, the model is able to identify key components within the system, offering stronger interpretability in economic terms.