Research on Photovoltaic Power Generation Based on Multi-dimensional Indicators and Models

As one of the world's largest energy consumers, the stability of China's power supply system is of great significance to the economy and society. In recent years, electricity demand has grown along with the rapid economic development, and although the power generation structure has diversified, coal-fired power generation still dominates, causing serious environmental problems. Renewable energy, especially photovoltaic power generation, has developed rapidly but also faces many challenges such as stability and cost. Therefore, it is of great significance to study issues such as power supply forecasting for photovoltaic power generation, site selection optimization for power stations, power generation increase and carbon emission impact.This study systematically analyzed the above key issues by constructing a multi-dimensional index system and multiple mathematical models. First, in the data preprocessing stage, the Kolmogorov-Smirnov test, box plot and other methods were used to handle outliers in the data; By using principal component analysis and t-SNE methods to reduce the dimensions of multi-dimensional indicators, a first-level indicator evaluation system including economic and industrial, energy consumption and structure, population and society, environment and emissions was constructed, laying the foundation for subsequent analysis.In terms of power supply forecasting, a cubic polynomial regression model was used. The results showed that the model had a good fitting effect, with an R² value of 0.9975, indicating that it could well explain the changes in power production. At the same time, ridge regression was used to solve the collinearity problem among the indicators, and a correlation model of carbon emissions with population, GDP, and energy consumption was established. In terms of photovoltaic power generation optimization, a model was constructed with the objective function of maximizing total power generation, and the particle swarm optimization algorithm was introduced for solution. By setting a series of assumptions such as initial installed capacity, growth rate of installed capacity, generation efficiency, etc., an optimized prediction of photovoltaic power generation for 2024–2060 was made, and the results showed that the algorithm could effectively increase power generation. In terms of carbon emissions prediction and optimization, carbon emissions and related data were collected and a multiple linear regression model was constructed to analyze the impact of photovoltaic power generation on carbon emissions . The projections show that with the introduction of new renewable energy sources, the upward trend of carbon emissions slowed significantly after 2020 and then dropped rapidly after peaking around 2030, with a significant reduction in total carbon emissions compared to the scenario without photovoltaic power generation.Overall, this study innovatively combines multiple approaches to validate the important role of photovoltaic power generation in achieving sustainable energy supply and supporting the "dual carbon" goals, providing data-driven decision support for regional energy planning. However, the model's adaptability to policy changes and market fluctuations is still limited and can be further improved in the future.