Effective analysis of the potential for electric energy substitution is significant for formulating development strategies and promoting local energy conservation and emission reduction. Analyzing the development trends of electric energy substitution under different scenarios can provide a scientific basis for regional planning. An improved particle swarm optimization-support vector machine model is proposed for predicting the potential of electric energy substitution under multiple scenarios. It analyzes indicators influencing electric energy substitution potential, such as the proportion of electricity consumption, energy consumption per unit of GDP, disposable income of urban residents, CO2 emissions per unit of GDP, and quantifies these indicators. Pearson correlation coefficient method is used to screen and introduce indicators into the prediction model. Four development scenarios—baseline development, technological progress, economic development, and low-carbon environmental protection are considered for predicting the potential for electric energy substitution. Actual data from a province in southern China is analyzed, comparing results with grey wolf optimizer-support vector machine（GWO-SVM）and SVM models, validating that the proposed method demonstrates good predictive performance. The electric energy substitution potential in 2030 and 2035 under various scenarios is also analyzed, providing theoretical support for future regional electric energy substitution planning.