The present paper puts forth a methodology for predicting the availability of piezoelectric high-entropy energy in road domains, predicated on traffic flow forecasting. The construction of a hybrid forecasting model integrating LSTM (Long Short-Term Memory Network) and particle swarm optimization (PSO) is predicated on the consideration of the spatiotemporal dynamic characteristics of traffic flow. The model incorporates time-period encoding, spatial interaction features, and an attention mechanism to effectively improve forecasting accuracy. To this end, experiments were conducted on the PeMS dataset, and the results show that the model achieves a minimum mean absolute percentage error (MAPE) of 7.19% in traffic flow prediction, thereby outperforming the standard LSTM model. The predicted traffic flow data is then utilized in conjunction with vehicle dynamics and piezoelectric material property formulas to estimate the piezoelectric energy availability. This method provides effective data support for the utilization of piezoelectric energy in intelligent transportation systems.