重大突破！中国超大电力系统转向国产求解器“天权”，“性能平均超越美国求解器14%”

Core Viewpoint - The article highlights the significant advancement of China's domestic solver software "Tianquan," which has replaced foreign software in the Southern Regional Power Market, marking a pivotal step in China's energy independence and optimization capabilities [1][9]. Group 1: Development and Performance of "Tianquan" - "Tianquan" is a domestically developed solver that outperforms its American counterpart, Gurobi, with a 14% faster calculation speed and a reduction in average computation time from 2111 seconds to 1827 seconds [4][6]. - The development of "Tianquan" faced initial challenges, with early tests showing it took 20,000 seconds to run a case, which was 11 times longer than the imported solver [6][7]. - After five years of development and optimization, "Tianquan" was officially launched in late 2022 and has since been used to support continuous settlement in the Southern Regional Power Market [7][9]. Group 2: Impact on the Power Market - The Southern Regional Power Market, which covers five provinces, has a daily trading volume of approximately 3.8 billion kilowatt-hours, surpassing the combined electricity consumption of the UK, France, and Germany [2][9]. - Since the implementation of "Tianquan," the market has seen a daily reduction in total generation costs by 70 million yuan and a decrease in carbon dioxide emissions by 18,600 tons [9]. - "Tianquan" has supported over 7,000 model nodes and 2 million clearing variables, facilitating significant milestones in China's power market reform, including the first province-wide and full-area settlements [9][10]. Group 3: Future Prospects - The successful deployment of "Tianquan" is expected to provide robust support for the national unified power market system, which aims to be partially established by 2025 and fully by 2030 [9]. - The solver is also being promoted across various sectors, enhancing production efficiency by 2 to 10 times and generating an average annual economic benefit of 26 billion yuan [10].